Column Bracing 4

[CTW]

I'm investigating existing columns (one's a crane column and the other is a building column).
The tips of the flanges of both columns face each other (y-axes are parallel). 6" x 1/2" plates are welded to the outside face of each flange on each side of the column, tying the two columns together. The plates are spaced 36" apart.

Can these plates be considered as bracing against buckling about the x-axis? I'm trying to determine the effective length.

 

[ChipB]

Best thing to do is:
1. Build the section in Autocad
2. Change the sections to regions
3. Do a "massprop" command and select your regions
4. Take note of the coordinates of the centroid
5. Move the regions from the coordinates of that centroid to 0,0 coord.
6. Do the mass properties again

It'll give you Area, Ix, Iy, rx, & ry. It's by the book after that.

 

[JedClampett]

I wouldn't. The bracing needs to be able to take a load (usually approximated as 2% of the column compression force) in the lateral direction. Unless you were talking about a small column being braced by a large column (which could take the load in weak axis bending), you're on pretty shaky ground. Also, if both columns are loaded, they could buckle in the same direction.

 

[JAE]

JedClampett is right. The plates only serve to require both columns to act together during any lateral distortion.

If column A is X feet long and, all by itself take 40 kips of axial load (based on its full unbraced length....

and column B is Y feet long and, all by itself take 60 kips of axial load (based on its full unbraced length...

and then you tie the two columns together with these plates, your combined columns can take 100 kips of load. The 100 kips can be applied to either column or in any combination as long as bearing and yield stresses are not developed. The combined column will not buckle until the full 100 kips is applied....

so you could have column A getting 50 kips and column B getting 20 and it would not buckle.

 

[andyfabian]

If I am understanding the situation correctly, there are two scenarios:

1. both columns are loaded and buckle simultaneously. In this case, I don't see how the plates could possibly help.

2. one column is loaded heavily and the other is lightly loaded (much below its buckling stress). The heavily loaded column is dependent on the lightly loaded column for lateral support. If the buckling force transfers through the plates and into the supporting column, the combined rx of both columns would have to be able to satisfy the buckling criteria. The Kl/rx ratio would not be based on the 36" unbraced length, but instead the floor-to-floor height.

 

[dlew]

CWT,

Here is my two cents worth of thought.

The 6" by 1/2" plates provide lateral support to each column about their weak axis from the crane girder down. The two columns are NOT braced about their strong axis by the plates. The critical buckling will be, then, about the strong axis (an advantage over not connecting the columns at all). Each column should take its own load, unless the plates are designed to transmit the horizontal shear (vertical in this case) between the two columns forming ONE composite section. Notice that if you consider the column as a composite section, there would be a moment due to the eccentricity of the resultan6t load of the two columns with respect to the c.g. of the section.

You could find a very detailed analysis of this type of construction in the AISC publication "Light and Heavy Industrial Buildings", by Fisher and Buettner

Hope this would help

AEF

 

[ChipB]

Ok, let me expand.

I would not include the plates in my analysis as they are not full length.

I did neglect to say to use the floor to floor height for "L" as Andy has pointed out.

I would check the combined axial and bending on the built up section due to the loading situations. You'll have a moment due to the eccentricity.

Think of back to back angle bracing. Their combined axial compression capacity is based on them being bolted together. Typically these bolts are spaced three to five feet o.c. From the description of the question, the scenario is similar. They are just using bars instead of bolts.

If you want to get in deeper, you can check the bars to make sure they are able to transfer forces. Calculate the differential compression between the two columns (P1L1/EA1 - P2L2/EA2) divide by the number of bars that you have, calculate what force is required for the bar to deflect that amount and then check the bending stress in the bar. Not entirely accurate, but it should get you reasonably close, and be conservative, without going into a FEA. Unless one column is extremely heavily loaded, I wouldn't do it.

 

[CTW]

Very Helpful! Thanks to everyone for their comments.

Now, would the K value on the shorter crane column be considered as 2.0 for the strong axis, rotation fixed and translation free (assuming the base is pinned)?

 

[haynewp]

The way I understand this which I believe is the same as dlew stated:
The columns are considered as "spaced columns" with this type of connection and cannot be analyzed as one composite member. The connections are not able to adequately transfer the shear between the 2 columns for it to be considered as one composite section, (as in laced columns, or composite steel beams with concrete slab and shear studs).
The columns are however able to share loads for buckling about the weak axis. This HELPS in forcing strong axis buckling control.(Still full unbraced length used for weak axis buckling, there is just the advantage of a sharing effect).
For strong axis buckling, the columns get no help from being tied together. The connector plates are theoretically hinged at each column, therefore there is no sharing effect for strong axis buckling and still full unbraced length for design of each individual column.

If the actual boundaries are as you described, then yes I would take KL=2x(full length) for checking the strong axis buckling of the shorter column.

 

[CTW]

haynewp, thanks for the insight.
I was hoping that no one would reply to my last post before I could post again and correct a mistake. I had written that K should be 2, but since this is a crane column, the girder would provide atleast a pinned connection at the top of the column (the girder sits on top of the column). So I would consider the K value as 1.0.

From what I've read, some consider the plates as providing bracing in the weak direction and others don't. I agree with everyone that the plates will not provide bracing in the strong direction and that each column should takes its own load. However, I would think that the plates do provide bracing in the weak direction, thus allowing a smaller effective length.

 

[haynewp]

No, I do not think you can analyze them as 2 separate columns braced at each connector for weak axis buckling. They will share the load due to interaction between the connectors and buckling will occur full unbraced length (ignoring the connectors as brace points).

 

[IJR]

It is really a long discussion and I have little to add, but I should agree with haynewp's last comment.

We usually analyze columns for buckling using simplified procedures which enable us to treat a column as singled out from other members. But this view is a result of simplifications from a detailed buckling (stability) analysis that include most of the time the whole frame or even the whole structure. One such detailed analysis involving a framed multistory structure would be simplified to a single frame buckling involving all beams and columns and further simplify to what books call subassemblage involving a single column and beams at the joint-the stuff we are familiar with and the Ga, Gb routine. In attempts to do this simplifications pals would first assume a good buckling shape of the whole structure or where they think fit they would do that for the representative frame.(Assuming shapes usually means attempts to solve an eigenvalue problem for a note)

Instead of thinking in terms of buckling I prefer therefore to think in terms of frame stability and where my mind can stretch further stability of a structure and buckling shapes.

respects
ijr

 

[dik]

6x1/2" plates @ 36" o/c is a pretty rigid connection and you may want to consider these as being 'rigidly' connected. It may be possible to use a filler bar and provide additional weld between the flanges. The new 'y-axis' will lie between the webs of the two columns and properties can be determined for the combined columns. This will give you a new kl/ry for a new stress capacity (ry=sqrt(Iy/A)). The kl/rx should not change significantly. Other combined section properties can be calculated as required.

It is likely that the original kl/ry governed the original design, so the improved one should permit you a higher stress level.

The varying loads on the two columns can be considered as an eccentric load giving rise to a beam column design.

It's possible to model the 'coupled column' and have the computer generate critical loading, but you have to be very careful in the modelling process.

 

[haynewp]

Using the plates as shear connectors sounds like batten column design to me. I don't know where CTW is from, but battened columns are usually not allowed in US design specifications. If you are going to assume a combined section, I would place a new diagnonal plate between the horizontal ties at each side to make it a laced member.

 

[IJR]

Thanks haynewp

I didnt know that battens are not allowed in US codes. I would like further insight. Do you want to go on tell us why(I follow US codes but I am not from US)

respects
IJR

 

[CTW]

haynewp-
could you expand on your statement, "battened columns are usually not allowed in US design specifications."
why they usually are not allowed and what spec's this comes from.

I am in the US.

 

[haynewp]

"Battened columns are generally not allowed in current U.S. design specifications but are in Canada" Design Criteria for Metal Structures.
"Open sides of compression members built up from plates or shapes shall be provided with lacing having tie plates at each end and intermediate points if lacing is interrupted". ASD 5-43 May also use continuous cover plates. 5-44

From my reading of AISC journals, batten columns encounter signfigant shear deformations which lead to a large decrease in axial capacity.

 

[dik]

I'd suggest that 6x1/2" plates would offer significant resistance against shear and flexural deformation. It might be an exercise to determine the shear force at the plate and check the moment generated by it and see what forces are required in the welds to resist this.

Some additional thoughts...

Lattice work that used to be used to tie columns together was generally light bar/strap material and attached by bolts or rivets... The shear deformation was such that only partial fixity could be assumed. Welding is generally a lot stiffer than bolting...

 

[haynewp]

What is the spacing between the columns and how heavily loaded are they?
Using the horizontal ties alone doesn't seem to be supported by AISC,I say proceed at your own risk.

 

[dbuzz]

What you are describing is essentially a battened compression member where two or more main components, battened together, act as a single member.

Section 6.4 of the Australian standard AS4100 "Steel structures" (

http://www.standards.com.au)

provides the guidance for the design of both laced and battened compression members, namely:
- design actions
- slenderness ratio limits of main components and battens
- slenderness ratio calculation for battened compression memeber
- effective length calculation of batten
- width and thickeness limits for battens