Checks Required at Concrete Column Transfer Slabs (Walking Columns) 5

[KootK]

Please review the sketch below (and attached). I am curious what others do to check slabs at column transfer locations. I'm thinking of "walking" columns in thinner slabs but the discussion is equally germane to full blown podium style transfer slabs as well.

The options:

1) Individual column punching shear with small moment.
2) Combined column punching shear with large moment.
3) Shear friction in the slab between columns.
4) Bearing stresses.
5) P x e moment in slab.
6) Vertical seismic acceleration effects.
7) One way shear between columns, similar to #3 (forgot to draw this)

I believe that the following checks are fairly standard: 1,4,5,6. Combined punching shear (#2) is a KootK invention as far as I know. Shear friction (#3) is standard practice at one of North Americas top tier structural firms. I feel that it's unnecessary as I can't imagine that ever governing over the diagonal shear failure modes (punching & potentially one way).

So, what do others check? And I'm not looking for FEM solutions. And I know it's awful.

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

 

[bookowski]

I don't know about #2. Similar to the other punching discussion I can see that in some cases it might be theoretically possible but it seems difficult to rationalize into something you can put on paper. At what distance do you consider this, 8x the slab thickness? always check but at some point becomes trivial?

By no FEM do you mean that you are getting all the demands by hand? Are you again assuming all pins? It seems that in some cases this would be unconservative.

#3 goes back to your does shear friction have to be satisfied discussion. I am on board that it always needs to be satisfied across any section that you can draw, it's just that at some point you can dismiss it by inspection.

Curious who the top tier firm is. My experience is that rating top tier by size/projects is not a good measure. I've peer reviewed a lot of jobs by those firms and often they are rubbish.

 

[KootK]

Thanks for joining in bookowski.

I don't know about #2. Similar to the other punching discussion I can see that in some cases it might be theoretically possible but it seems difficult to rationalize into something you can put on paper. At what distance do you consider this, 8x the slab thickness? always check but at some point becomes trivial?

I think that this is a completely different animal. In the other discussion, the effect was secondary and not required for equilibrium. Here the effect is primary and absolutely required for equilibrium. In the section view of #2 above, I showed a section of the slab rotating under the effect of P x e. Something needs to prevent that from happening. As I see it, the choices are:

1) Fixity between the columns and the slab. Due to detailing limitations, it seems sketchy to count on that.
2) Bending forces applied to the perimeter of the slab section. Due to the small area involved, flexural capacity will be insufficient.
3) Eccentric punching shear. I believe this to be the dominant contributor.

I've already rationalized it into something that can be put on paper. In the plan view of #2 above, I've defined the punching shear perimeter explicitly. Other than the odd shape of the punching perimeter, this would be the same as a regular eccentric punching shear check.

As I mentioned, I'm considering a case where the columns are in close proximity and the chunk of slab that punches would be fairly rigid. Once the the distance between columns becomes large, slab flexibility would kick in and a punching shear model would no longer make sense. Is that distance 8 x slab thickness? More? Less? Dunno. We're off the reservation here.

By no FEM do you mean that you are getting all the demands by hand? Are you again assuming all pins? It seems that in some cases this would be unconservative

All that I meant is that I don't want anyone answering my question with "just run it in SAFE". I'm looking for fundamental conceptual understanding here, not pass/fail answers for a particular project.

Curious who the top tier firm is. My experience is that rating top tier by size/projects is not a good measure. I've peer reviewed a lot of jobs by those firms and often they are rubbish.

I agree that big <> good. I think that big firms often become cultish in terms of their engineering dogma. That can lead to a lot of blind procedure following and not a lot of critical thinking. That being said, I was formerly an employee of the company that I mentioned. Their bread and butter is high-rise residential condos. They'll do a few hundred each year and almost all have transfer slabs near grade and near the penthouses. I can say with confidence that;

1) There have been no reported failures. Normally I don't put much stock in that argument. However, in this case, the predominant load is self weight. These column transfers are effectively load tested to a degree that most things aren't.

2) People at this company have spent a lot of time thinking about this. I think that their out to lunch about the shear friction business for reasons that you and I have discussed at length in the past. However, I definitely think that they're are on to something with the check. I just think that the check should be one way Vc, not shear friction. I'll elaborate on that in my next comment.

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

 

[WARose]

For #2, I assume you refer to the shear increase from moment transfer from the slab to the column. A good reference on that is ACI 352.1 R-89. They've got some examples and it includes a discussion on proper reinforcement details and the effect of openings in the slab.

One thing about the loading that should be used for design: is watch out for pattern loading. I remember checking one some years back where a pattern loading (for the live load) actually controlled for design.

 

[KootK]

#3 goes back to your does shear friction have to be satisfied discussion. I am on board that it always needs to be satisfied across any section that you can draw, it's just that at some point you can dismiss it by inspection.

It is actually #7 that worries me the most. It's the same as #3 but one-way diagonal tension shear (Vc) rather than shear friction. See the attached sketch. When the columns are spaced further apart, the shear stresses have time/space to shift around. In that case, I think that punching checks on the individual columns makes sense. When the columns are close, however, my gut tells me that the bulk of the shear makes its way across the slab via localized one way shear.

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

 

[KootK]

Thanks for your response Rose.

For #2, I assume you refer to the shear increase from moment transfer from the slab to the column. A good reference on that is ACI 352.1 R-89. They've got some examples and it includes a discussion on proper reinforcement details and the effect of openings in the slab.

Well... yes. What I have suggested is actually quite unconventional however. In fact, I've never seen it proposed anywhere. It's a punching shear perimeter that encompasses two, offset columns and transfers P x e moment into the slab. Procedures for checking eccentric punching shear on a given perimeter are fairly well established. The big question here is whether or not the check is required.

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

 

[bookowski]

By 1) are you saying that you assume no fixity between the slab and column? Typically if this is a transfer slab it's relatively thick and the typical detail would include matching dowels up for the column, I wouldn't consider that a pin. That also means that in your punching shear checks in general you have no unbalanced moment which makes up a good majority of the demand.

A few hundred high rise flat plate residential in N. America per year? Not many people doing that. Cantor/wsp, tt...?

 

[WARose]

It's a punching shear perimeter that encompasses two, offset columns and transfers P x e moment into the slab.

If it was me, I would only consider the perimeter provided by the one column (on the bottom). The one on top is loading the slab/column.

 

[KootK]

By 1) are you saying that you assume no fixity between the slab and column? Typically if this is a transfer slab it's relatively thick and the typical detail would include matching dowels up for the column, I wouldn't consider that a pin. That also means that in your punching shear checks in general you have no unbalanced moment which makes up a good majority of the demand.

In a real design of a thick transfer slab, I generally would consider fixity. For this discussion, however, I have assumed pinned connections for the sake of simplicity. Assuming fixity in this discussion introduces new forces which muddle things and don't materially affect the arguments being proposed. That's my opinion at least. If I'm wrong about that, I'm interested to hear more.

A few hundred high rise flat plate residential in N. America per year? Not many people doing that. Cantor/wsp, tt...?

I shall not tell as I have questioned their methods and do not wish to be guilty of sullying their reputation when they're not here to defend themselves. The hundreds of high rises they do aren't Burj Dubai's. The overwhelming majority would fall in the 15-30 story range with the odd 60+ thrown in there. Maybe mid-rise would have been more apt. Regardless, oodles of multi-story concrete with transfer slabs in the mix.

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

 

[KootK]

If it was me, I would only consider the perimeter provided by the one column (on the bottom). The one on top is loading the slab/column.

This has me worried WARose. At minimum, the upper column needs to have an individual column check to ensure that it won't punch downards through the slab. Take my sketch, flip it upside down, and take another look at the situation. I'll think that you'll come to see that both columns require punching shear checks.

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

 

[WARose]

Both do require punching shear checks (one for each).....but I thought the combined perimeter was non-conservative.

 

[jayrod12]

KootK, For #2. As you've sketched it, wouldn't a quick initial check to see if it's a concern be P*e/(l*b*d) where l is the overall length of the piece of slab from failure line to failure line, b is the smaller length of failure line (next to the smallest column) and d is slab depth. This would give you a shearing stress on that face which you could compare to vc. If it's close then a more accurate analysis may be in order, if it's not than no further check is required.

That calc can't take much more than 5 minutes and would put you at ease (or on alert)

 

[KootK]

Ah... now I get it WARose. I wasn't suggesting the combined perimeter instead of the individual column punching shear checks. Rather, I was proposing that the combined perimeter might need to be looked at in addition to the individual column punching shear checks. The combined perimeter is relatively long but the moment that it needs to resist may be very large (P x e).

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

 

[jayrod12]

The difficulty I have is how to account for the fact that the direction of the shearing stress switches somewhere along the perimeter. I see the hunk of slab as more of a beam trying to rotate about it's centroid then I do as a lump of concrete getting punched out.

 

[bookowski]

That was just a range finding shot, I aimed too high. Hundreds of mid rise in N. America per year still means it's likely a ny firm, a couple of tiers down from the original guesses - that puts you in the dce, gace, rosenwasser level..

My gut impression is that you can not assume pins and ignore the fixity. There is a lot going on from slab moments + transfer moment and the unbalanced portion will eat up a good portion of your capacity.

The check should be for max stress occurring anywhere and verifying that vu < phi*vc (little v's). If you take your original #1 and #2 plan diagram and look for example at the far left perimeter (up/down on page line) the stress is whatever the stress is, it doesn't change by how you draw your critical sections. By checking the two different sections as you've shown my guess is that you'll end with quite different answers, which means that one is not correct. It seems like this is taking a complicated indeterminate system and drawing two simplified determinate systems and saying which one is correct, I'm not sure either one is. I think you need to include the unbalanced moments and then look at the stresses d/2 away.

 

[KootK]

I think that's a clever and practical proposal Jayrod. Thanks for suggesting it. Would you use one way shear stress limits or two way? They are different for reasons which have never been entirely clear to me. Another quick and dirty might be a two way check on a rectangle encompassing both columns and approximating the real perimeter, whatever that is.

The stress in your check would also need to include something to reflect that plane's contribution to resisting the axial load "P" though, right? P/4L if it were a square column?



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

 

[KootK]

I agree with your last post in its entirety Bookowski. However, you're getting too fancy on me. My primary concern isn't how #2 ought to be checked; rather, it's whether #2 needs be checked. I've never seen the combined, two column shear perimeter proposed before anywhere: not in any text book or design guide, not in any company's design standards, not by any of my colleagues. Your last post makes it sound as though you think #2 should be checked in some fashion. If that's correct, it's a big step in its own right.

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

 

[KootK]

The company that I mentioned did a study on transfer slab thicknesses to get a feel for the level of consistency from one designer to the next. There was none. For similar conditions, one engineer might use a 900 slab while another would use a 1200 slab. Much of the scatter seemed to have come from differences in what various engineers thought needed to be checked. It was very interesting.

I'd still like to get some thoughts on #7 and my last sketch. If that's a valid failure mode, the rest is just window dressing.

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

 

[bookowski]

Shear stress needs to be checked around the entire area, after that it does devolve into a how. If there is a way to accurately model/calculate the unbalanced moments and axial loads then you could look at the d/2 stresses away. The big vs little section is a how argument isn't it, it's how do you get what the true stresses are (since it seems like we agree that they are what they are, #1 and #2 are two guesses at 'how' you find them - only one is accurate).

I agree with one way shear entirely and that it's a little bit fuzzy how far away you have to be before you spread that out. It is not the same but similar to a question I asked recently about 1 way shear in a mat foundation. Several responses opined that it's impossible and only the full mat width needs to be checked but I disagree.

 

[jayrod12]

For #2: I would be inclined to use two way shear resistances as it seems most applicable to the shape of the chunk of concrete. I guess yes you should be including a portion of stress due to the axial load in the column (although technically you're accounting for it in the moment). It would take a couple trial runs for me to determine what to account for and what not to. I think checking the resistance against the moment alone, and also checking the single punching shear of one column only would tell me if I need to worry about both at the same time.

As for number 7. How do you determine the length of the one way shear failure plane? I think it is likely that it may start failing in one way shear directly between the columns, but once the cracking reaches the edges of the columns it would need to turn into two way shear. There is no way it is going to cause one-way shear for the whole length of your slab.