Modeling Cocnrete Column Joints

[hippo11]

We have a situation where the concrete column below is going to be smaller than the concrete column above, and the column centroids will be eccentric.

What program would you recommend to model this beam-column joint so we can see how the stresses flow betweent the columns, beams, and slabs?

Thanks.

 

[JAE]

First off...why go smaller below? Seems sort of wrong.

 

[hippo11]

Well, we have to maintain clearance for the drive aisle below and we have to maintain residential layouts above. And we can keep the slab above at 8" thick for 40+ floors instead of 10" thick if we shorten the span by having that larger column above.

It's a long story, but that's what we're doing.

Thanks.

 

[JAE]

Beam and slab design? or concrete joists? Anyway - assuming you've got a concrete beam in line with the columns - your column eccentricity would impose a moment at the column joints that would affect the beam in terms of moment - but probably not shear. Since ACI allows you to ignore concentrated loads closer than d from the face of support, your beam shear may not be affected other than that caused by the moment at each end.

In a sense, you've perhaps got a sort of corbel situation where some extra reinforcing per ACI 11.9 is warranted.

 

[hippo11]

Beam and slab design.

And yes, we have looked at the ACI provision saying that you can ignore concentrated loads closer than "d" to the face of the column, and in our case we have an offset larger than "d". ACI allows you to do this because loads closer than "d" will transfer directly to the support using strut/corbel action and not engage shear action in the beam...And since in our case we're farther away than "d" we'd like to analyze the joint to see quantitatively how the loads are transferring...exactly how the shear is flowing at these joints, i.e., how much shear is flowing to the column below and how much is flowing to the beam...

I wanted to see what program people thought would be best to use to model this joint...

 

[JAE]

Keep in mind that ACI 11.1.3.1 does not say you IGNORE the concentrated load. That is a mis-reading of the text.

What it does say is that when designing for shear in the beam, you can use the Vu located at d from the face for the section of beam between the column face and the point at d.

The point load is still applied, and still induces shear into the beam along its length, and in your case does induce some additional moment into the beam.

As far as modeling goes, we have a program that allows you to make a rigid end at whatever distance out from the centerline of the column. This simply adjusts the stiffness matrix and does alter the flow of load between column and beam. If you don't have a program that specifically does this, you could simply add joints at the face of each column (along the beam length) and add joints along the column length at the top and bottom surface of the beam, and then make the small segments there (four of them) a higher rigidity to better model the non-flexural condition of the beam/column mass of concrete at the joint.

 

[hippo11]

Forgive me for not being more clear in my posting....

When I said "ignore", I meant with respect to shear design of the beam, i.e., take Vu at "d" from the support as opposed to right at the support. I understand the code is saying that the point load still exists within "d" and contributes the the forces that the structure is experiencing, but that you are allowed to take the Vu at "d" from the support.

And also of course you can't do this if the support is in tension or if the applied point load is toward to bottom of the supported member.

But I'm not trying to get caught up in the semantics of the postings.

I was looking for recommendations on programs to make an FE model, not a frame member and node type program, if that makes sense.

This way I could see the shear stress flow across finite elements, using contour plots, from the top of the member in question to the bottom of the member, and check if there's any shear stress > about phi*10 sqrt fc' (ACI 11.5.6.9) or so, maybe less, because this is about the most shear a beam with stirrups could handle even if you really pack in the stirrups. This will tell me if my beam cross section is reasonable.

We'd like to really look at the stresses instead of just invoking the ACI provision which seems most fitting.

I think I'm leaning toward ETABS, and I was wondering if you guys thought this was a good FE program to use or if there's another FE program that would be more appropriate.

Thanks.

 

[JAE]

hippo11 - sorry, I don't have any FEA recommendations for you on this...just that it would have to be a 3D element I would think.

But I'd also just throw in this thought: do you really need to micro-analyze this? It seems like, for one joint in one building, you could use some rough assumptions and go a bit conservative in your design and avoid the software cost and time to build the model and analyze it....just my 2 cents. Concrete is an amazingly malleable and forgiving material if you reinforce for all possible stresses.

 

[hippo11]

True, but we're actually looking not at just one joint,but at joints in floors 4 through 15 of a 62 story building, and the cross section is jumping back 18 to 24 inches or so at each level, so I don't know what kind of rough assumptions I could use in this case...what did you have in mind? I don't know how else to check to see if my beam cross section is okay.

We've already got ETABS and SAP and some other programs so that's a sunk cost.

Thanks...

 

[JAE]

I don't have a perfect picture of the condition, but if you generally have a larger column bearing down on a smaller column, with a concrete beam going through the joint, I would think that what would happen is that the stress in the larger column would stream down to the 18" as it approached the joint and there would be a surrounding set of four wedges at the bottom of the upper, larger column, at each corner, that would have no stress, or would have tensile stresses that might cause a lateral blow-out of the corners.

With the possible eccentricity you speak of, and the fact that the beam sort of invades through the whole thing, I would imagine the stress flow simply begins to neck down to the 18" from the 24 to the 18, through the beam. Again, this is really just a compressive narrowing of the stress field as it heads downward. I can see just assuming a corbel-type condition and adding reinforcing in the joint to accound for this "imaginary" corbel.

I know I'm a bit vague here...but just don't know the exact geometry of your condition.

 

[hippo11]

That helps, I think I follow you.

Our beam is 27" deep (limited by car clearances from below), and when we have offsets of 24" from column above to column below, we're hoping that since "d" of the beam is about 24 or 25 inches, that ALL the compressive stress will narrow itself, as you say, and flow into the smaller column below.

But since we're right on the edge of being "d" away from the support, we're suspicious that some stress might shed away instead to the beam as beam shear.

That makes sense to look at it as a corbel, I guess the question is, if the stress from the column does shed out to the corbel, how much? And how wide will that corbel need to be, and how many stirrups?

But here it would seem that your stirrups would be vertical, not horizontal as you would do for a corbel.

Thanks.

 

[JAE]

You could detail a "corbel within a beam" where you'd have vertical and horiztonal stirrups/ties.

As far as how much the load sheds out through the beam, I wonder if you could first try to make an assumption - perhaps even just visually, and then adjust it up as a safety factor to be conservative.

You know that the amount of shear flow through the beam is between 0% and 100%. So if you visualize that 30% goes through the beam and 70% necks down directly into the lower column, then use 60% for the beam/corbel design. I know this sounds sort of rough, but keep in mind that force follows stiffness. The column below most likely has much higher stiffness so the majority of the load would zip right to it, as opposed to following the beam path which is longer and "softer".