Concrete Column Walk

[bookowski]

Question regarding a shift in plan location and/or orientation of a concrete column, "column walk".

In my training I was taught to check the following: confirm that the overlapping area works as a short column (x1 by y1 in attached sketch), and that the resultant overturning forces from the column P x e can be resolved in the slab diaphragm with addition reinf as applicable - this is more or less saying to do a strut and tie although at the time no one called it that. It seems like there may be more to this but I am not sure what is missing.

In the attached sketch a column rotates in plan such that an eccentricity would be created in two directions, is resolving the Pex and Pey force couple along with checking the overlapping area sufficient? With these checks alone you can get pretty acrobatic - it seems like I am missing some other limitation.

 

[KootK]

I was taught exactly the same thing. That short column restrained every which way works pretty well don't it?

The trouble, in my mind, is the assumption that you only have to design "a" load path, even if you don't know "the" load path. Saying that all of the moment can be resolved in the slab, and that none of it will adversely affect the columns, seems naive to me. To cover all bases, I would do this, in addition to what you (we) were taught:

1) Make sure that punching shear in the slab is also designed for P x e + regular slab moment transfer.

2) Make sure that your columns are also designed for P x e distributed to the columns above and below.

I've found that strut at tie models that connect the centrelines of columns rarely work. Slabs are generally proportioned thin enough that these struts become too steep.

The punching shear check can be a weird thing. Is it based on the lower column, upper column, the overlap area, or some other hybrid? That takes some some judgement in a lot of cases. It would be conservative to say the overlap area but that can be pretty punitive.



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

 

[KootK]

3) Depending on how you choose to look at things, P x e in the columns may result in in-slab lateral ties/strut at the floors above and below the walk. Similar to how you might tie a sloping column when it straightens out. Usually pretty minor forces.

A question that I intend to submit one of these days is this: what checks are required when the "walk" results in columns with no over lap at all.

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

 

[asixth]

I have let the slab diaphragms be the elements that rectify the horizontal component that comes from eccentricity, reinforced the slab to accounted for the additional forces that arises from the small transfer, and relied on the whole column length for the loads to transition from the "above" cross-section to the "below".

For highly stressed columns I have provided deep slab thickenings to transition the column loads thru the slabs, sometimes up to 1000mm thick.

 

[bookowski]

Sounds like the methodology is pretty consistent. I'm still not convinced that there isn't more to it.

Kootk - I agree with checking the column for P x e moment, although it's probably a little bit conservative. I'm not sure about the punching shear one. How that P x e goes into a slab to column connection isn't clear to me, I'm sure there is something going on but it doesn't seem very clear what exactly. In general I have issues with punching shear that are stronger than your shear friction issues so I won't delve into that one. I do agree that the punching perimeter is a bit fuzzy but using the column below seems defensible.

 

[KootK]

I agree with checking the column for P x e moment, although it's probably a little bit conservative.

I'm not sure that we're talking about the same P x e. I do, however, get the sense that we are all talking about the same fundamental concept. Please review the sketch below. Consider these two design checks:

1) Checking the little mini-column within the slab overlap area.
2) Checking P x e in the columns as I've shown below.

I propose that check #1 implies check #2. You need both for a consistent design. I've heard it proposed that the column should be designed as a true strut and tie element. You know, a vertical truss with ties as webs etc. I don't do that... mostly 'cause it's hard. Instead, I add the moment and shear diagrams shown below to the other column effects and then design it as I would normally design a column (sectional method).

I consider the model shown below to be the most likely load path. It ought to be the stiffest load path because it involves primarily axial load transfer. As a result, I'd like to change my answer a bit regarding what needs to be checked:

1) Mini-column
2) P x e and extra shear in columns.
3) Horizontal strut and tie represented by floor slabs.

Somewhat surprisingly, I don't think that anything needs to be done with the slab flexural reinforcing. It's not required for equilibrium and, of all of the failure modes involved, that's the one that's probably the most ductile and insensitive to overload. I'm still inclined to do something about punching shear but I'm not sure what. That also is not required for equilibrium but is a brittle failure mode sensitive to overload. I'd hate to see the column rotate a few degrees and punch through the slab. Maybe punching shear resistance could be provides via stud rails at these locations to provide some ductile capacity.

I have some strong opinions on punching shear myself. If you start a thread, I'll show up.

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

 

[KootK]

I guess this suggests two more -- likely non-critical -- failure modes. Shear friction at the ends of the columns and strain in the slab tie bars. If the nearest shear wall is seven miles away, 2-15M might not be the way to go.

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

 

[bookowski]

Yep, I agree with the moment check. Although that moment diagram assumes pinned ends on everything. If you had some moment continuity across the walk you'd cut that in 1/2, and if you had the continuity of a column above you'd reduce it further. That would be geometry specific and require some judgments, i.e. 4" overlap probably a pin, 24" overlap probably not.

Agree about flexural steel as well, additional tension steel is perhaps reqd. in some cases in the slab but not flexural.

Also agree that theoretically there is a direct shear plane there. Unlikely to be a concern but theoretically it exists.

I'm still not convinced that you could ever rationalize anything worth adding into the punching shear checks.

 

[KootK]

Also agree that theoretically there is a direct shear plane there. Unlikely to be a concern but theoretically it exists.

You know I can't resist a plug for shear friction...

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

 

[KootK]

http://i57.tinypic.com/4se0qp.jpg

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

 

[KootK]

Although that moment diagram assumes pinned ends on everything. If you had some moment continuity across the walk you'd cut that in 1/2, and if you had the continuity of a column above you'd reduce it further. That would be geometry specific and require some judgments, i.e. 4" overlap probably a pin, 24" overlap probably not.

I get what you're saying and I agree with the physical phenomenon that you've described. That being said, I don't think that it's the way to go procedurally:

1) "Knowing" the column moment is impractical. I think that you want a conservative estimate.

2) Analytically, I want the issue dealt with between the two slabs that bound the walking column. That means the diagram that I posted above where the struts meet up with the centrelines of the the columns above and below. Assuming moment sharing to the columns above and below equates to the struts hitting the columns above and below eccentrically. Then you just have to deal with the same issue for two more floors... and two more after that, until the struts finally coincide with column centrelines.

I'm still not convinced that you could ever rationalize anything worth adding into the punching shear checks

I mostly agree. The sketch below shows where I think a punching shear failure could originate from. So long as the strain in the slab horizontal ties was kept small enough, this probably isn't an issue.

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

 

[bookowski]

Agreed - that's why I said that I agree with designing for the full moment but it's likely a bit conservative.

I agree that there is some interaction with the punching failure mode but I doubt that what you can come up with to integrate into your check is accurate enough to be worthwhile. You would need to estimate that imposed rotation somehow with a full model of the slabs above and below. In some cases the rotation could actually help you by relieving unbalanced moment.

 

[KootK]

Substantial consensus... reached! This has been great for clarifying my thinking on this. Sometime soon we'll have to tackle walking columns that don't have any overlap. Scarier still.

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

 

[DaveAtkins]

Sorry to foul up your substantial consensus, KootK

I have been thinking about your first sketch, and I think one can argue there is no moment. Since there is a direct load path (even though it is slightly skewed), I would check if the columns can be shown to act as "leaner columns." The vertical force and shear force are resolved to be a simple axial force along the skewed axis of the column.

DaveAtkins

 

[KootK]

I have been thinking about your first sketch, and I think one can argue there is no moment. Since there is a direct load path (even though it is slightly skewed), I would check if the columns can be shown to act as "leaner columns." The vertical force and shear force are resolved to be a simple axial force along the skewed axis of the column.

I think that this would only result in no column moment if the columns actually did lean. I held the same opinion myself until I worked out the FBD at the top right of my first sketch. Can you work out an alternate FBD that supports your theory Dave?

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

 

[DaveAtkins]

You are correct, KootK. For my approach to be valid, you must assume the column is skewed (by ignoring the portions of the column that are not a part of the direct load path). I think it would be best to include the moment in calculations--however, I think the column may actually behave as a leaner column.

DaveAtkins

 

[bookowski]

This is pretty much in agreement with above discussions, the moment falls somewhere between 0 <= M <= P x e.

This relates to a recurring assumption/problem which is can you ignore a part of a member/structure that you don't need, i.e. can you pretend that the column is leaning and the extra parts are extra that you didn't need. Doesn't make sense from a stiffness/compatibility view, but likely goes in the wash with our factors of safety most of the time.

 

[KootK]

I absolutely agree that, if you tested the column or modelled it with FEM, you would find a diagonal compressive stress field that reflects Dave's leaner column theory. And, if things went south and the non-leaner column began to fail in flexure, I suspect that an unreinforced leaner column would develop and would be one of the final stages that the system would have to pass through before collapse.

That being said, I believe bookowski's last comment to be wholly accurate. Until extreme things start to happen, the non-leaner parts of the column cross section will behave compositely with the leaner column and stresses and strains within the column will reflect that. Hence our consensus (fingers crossed) that some level of P x e moment ought to be considered.

I’ve seen colleagues install leaning rebar cages within plumb walking columns. I get the impetus but feel that it’s a mistake for several reasons:

1) Constructability.
2) It actually makes the columns less effective at doing their day jobs: supporting slabs via flexure and punching shear.
3) Design complexity. Once you ignore parts of the column for axial load, it seams logical that you should do the same for flexure and punching shear. Ick.


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