2 beams crossing without a column

[cesaramorim]

Hi, this might be a dumb question but i'll go for it anyways.

If I have 2 beams that are crossed in a corner, for example like this:

And they don't have a column there to support them, what kind of behaviour should I expect on the cross node?

I've made a simple fem (that i will try to perform in sap2000 too) and the result is this:

It's very different to have 2 beams, than a single one, so what could be meaningful to ask here, is, why can't I consider it a single beam with a change of direction?

Thanks for any help

 

[JoshPlumSE]

I think the key is that you want them to behave as a single member with a bend in the middle. Therefore, you must make sure that they connect to each other in a way that allows it to act continuously. That means a moment connection of some kind.

 

[cesaramorim]

Thanks JoshPlum, I was pretty sure it was some sort of dumb question, but it's a valid one, and I think sap2000 is giving me different deflections than that model, but i'll develop it a little further to make sure before posting here. I'll take a look on the info you provided.

 

[KootK]

why can't I consider it a single beam with a change of direction?

You can consider it a single beam with a change in direction if you wish to. The reason that this is not generally done is that it generally makes for expensive and inefficient construction. Consider:

1) Your beams would carry a bunch of torsion. Most steel sections are terrible for torsion.
2) You would need moment / torsion connections at the beam intersection. Those will be substantially more expensive to fabricate compared to simple shear connections, particularly if weld testing is required.
3) Your analysis will be much more complex, and much less predictable, with torsional continuity included.
4) If you make the dog leg connection in the shop, the member may be a hassle to ship and erect.




I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[racookpe1978]

But your dimensions and angles (the bent beams shown in the model) do not match the 90 degree corners of geometry in the drawing.

I'm surprised you think a model of one geometry can predict the stresses and strains of a different geometry. Rather, you'd have a vertical point load coming down the vertical member into two cantilevered horizontal support beams. The stress model would need to show both ends of the two horizontal beams, their single connection under the vertical, and the supports or anchors of the other two ends.

 

[FoxSE14]

Find the nearest column to the corner. Can you utilize (or add) yet another column nearby to support a backspan? -say a distance "L" away?? See crude cartoon model below



______________________V
^............"L"............^.......

(ignore the dots, the board wouldn't allow blank spaces easily).
V -- represents the reaction from the other beam at 90-degrees.
^ -- represents support locations

 

[jwilki]

Agree. This should be a straight forward cantilever with backspan. So Cantilever the short direction (over the green column) with a back span, then all is required is a simple connection to the other (longer) beam. Hide in roof space or wall frame to suit.

 

[cesaramorim]

Well the sap2000 model provides exactly the same results. What I don't like there is the deflection there. In the model I've shown the node transmits all the stresses to each bar, (it's not a articulation, bending moment is transfered as well) But because the model considers it two bars there's a big deflection. And I have a cantilever right ahead! So backspaning it's really not something I'm considering for now and placing a column in the node, or in nearby is pretty inconvenient because I have two parking lots below with a different configuration from eachother.

Here you can see the 2 adjacent columns and the cantilever ahead (varanda in most cases means a cantilever)

@racookpe1978 I didn't think it could, just tried it anyways Do you know any literature on this?

A few notes, they will be concrete elements, so @Kootk any problems with torsion or connections aren't problematic as long as I can get the real deflection of that node. I don't think a concrete element will behave like 2 different bars. However I have no idea how to analyze it. Is there any kind of literature on this?

 

[FoxSE14]

I've lost track of your original question. Based on what I think you're getting at, I recommend:

1. a. If I assume that north is up on the screen - is there another column farther south, but in line with, the green one? (@ an adjacent frame? @ an opposite side of the building?)
1. b. If yes, can a reasonably sized beam span from this column to the green one?
2. IF the answers to 1a. and 1b. are both yes: why not run the rebar through the column joint at the green column, and all the way to the end to support the zig-zagging concrete beam?
3. When they build your zig-zagging concrete beam in the real world, will they run the rebar continuously along its length? If so, then your model needs to ensure there is full fixity at the zig-zag joint.

It is entirely possible that there are other conditions on your project which make any of the recommendations we give you quite trash-worthy; regardless of your path, make sure you enlist the assistance of a more experienced coworker, supervisor, or senior engineer before you get too deep into the weeds.

Get familiar with your software and be smarter than it is. Make sure you are modeling your joints/boundary conditions to reflect the real world. The software will not do the job for you. If you are getting deflections you do not like, you either have modeled boundary conditions/something else incorrectly, you need larger members, you have incorrect loads, etc. Good luck.

 

[Teguci]

This is reading more like a cantilevered flat slab design rather than a beam slab design. Sort of like Frank Lloyd Wright's Falling Water (except this time the engineer is going to design the cantilevered slabs correctly).

Have you looked at thickening the slab or even maybe doing a P/T design? It should make your formwork and rebar detailing a lot simpler and provide accurate modeling for what the structure wants to do. I'd still keep the column to column beams, but I'd change the cantilevered corner beams to a single diagonal beam from green to red column.