Column Bracing to Increase Column Capacity

[NinerStruct]

This came up in the thread765-368271, where existing columns were noticed to be under-designed. In discussing how to remedy the issue, I was looking to simply decrease the braced length of the columns by adding a bolted T-C brace in each axis, angled at 45 from the column up to the roof structure. In most cases, I only have access to two sides of each column due to existing ductwork, which is why I only had one in each axis. My concern with this approach was that the brace would be adding an additional horizontal load on the column from the roof structure as the existing roof beam deflected under snow load, and actually cause more harm than good when I add the moment and have to start using the interaction equations.

I wanted to post the question in this forum to get others' opinions on this as well.

Thanks

 

[bookowski]

Seems like you'd have to model it and run the numbers. Since it will only feel load from snow (or live) loadings, and assuming that the brace point is relatively far from the mid-span of the beam I would guess that the imposed load is small - but without running the numbers I don't think you can know for sure.

 

[IFRs]

Bracing prevents translation, to keep the axial force concentric. If the bracing bows the column then it's axial load becomes eccentric. You may need to see how stiff the roof is, how it deflects and thence how much the bracing bows the column.

 

[TehMightyEngineer]

You wouldn't happen to have a drawing or sketch of the column?

What about welding on reinforcement plates to the column?

Maine EIT, Civil/Structural.

 

[spats]

Adding braces is not necessarily a good idea. You need to make sure the braces are strong enough and stiff enough to force an inflection point, and that also involves the strength and stiffness of the roof members you're bracing to. There is a good paper on this subject by Joseph Yura.

 

[NinerStruct]

Here's a link to a quick plan sketch and elevation of the columns. This is in an existing building and the columns are buried in l.g. demising walls, so I was trying to minimize demolition and interruption of use at the building. The easiest way to do that was reducing the braced length with a brace in all 4 directions. But with ductwork and other utilities, I only have access to two sides of most columns so I was looking at a T-C brace in two directions.

I've begun to do a quick Risa 3D model, in which I add the physical brace and record it's axial load from only snow. I then convert that to it's horizontal and vertical components and add them as joint loads for snow only at the brace elevation, turn off the physical brace and set the braced length of the column in the properties. Does this make sense? Otherwise, the model imposes dead load horizontally on the column as well.

TehMightyEngineer, I had thought about increasing the radius of gyration by adding tubes to each face, but wanted to explore the braces first, as it would be less material and less demolition. That's my next exploration though, and how to figure exactly how much of each column needs reinforced.

 

[bookowski]

For simplicity I would model the column, beam(s), and brace and look at it under snow load only. Apply the dead load as a point load to the t.o.column. That way you don't need to mess around with resolving components by hand and you also capture that the brace does not have a rigid end support, the column has some flexibility (should reduce the axial force in the brace).

 

[TehMightyEngineer]

What about filling the column with concrete?

Maine EIT, Civil/Structural.

 

[paddingtongreen]

I don't see it.

The columns are loaded now and are assumed to already be in single curvature since they are not blemish free. Adding a brace now might force them into "S" curves under additional loads (roof or snow), or reinforce the existing single curvature, but either way, snow load will also put a diagonal force into the column from the roof level, and push the column back into single curvature. The vertical component of the brace will be some of the load that would otherwise reach the column from the roof system.

You cannot add the brace and say it is effective in bracing the column and then say it is not effective in bending the column.

Michael.
"Science adjusts its views based on what's observed. Faith is the denial of observation so that belief can be preserved." ~ Tim Minchin

 

[hokie66]

I commented in the Ethics thread, and am glad to see further comment here. I think the diagonal braces can be effective in reducing the unbraced length. Bending has to be considered, but based on the relative stiffness of the column and beam, I doubt bending will be much of an issue. In the other direction, with bar joists, that may be a bigger deal, and the connection to the bar joists would be more of a problem.

 

[haynewp]

I would add a new section to the column at each side within the wall full height. The new built-up column will have increased strength against buckling in the other direction as well, without adding to those faces. Assuming you can't add a new section on all faces.

 

[steellion]

I think bracing the column to reduce the unbraced length is a viable option. Yes, the brace will add a horizontal force to the column that will need to be checked in combination with the other column loads, but the benefit of reducing Lb may counteract the negative and get the column to check as acceptable. You wouldn't need to put the roof dead load into the kicker because you are adding a kicker to a roof that is already stressed by the dead load. The kicker would transfer only live/snow loads to the column.

 

[haynewp]

In the other direction, the brace would be supporting the joists under snow load. This will change the moment diagram of the joists and could cause stress reversals in the joist members.

 

[NinerStruct]

I appreciate everyone's input. It's brought up a number of good points.

I think bracing in the direction of the joists is more complicated than I initially thought. I hadn't thought as much about bracing effect on the joists, only that they wouldn't put as much load in the brace as the beams would, and that was my main initial concern.

I know that there are cases where adding the braces would be beneficial, but the more I look at it, I just don't think I have the right scenario to adequately implement this type of fix. I think that the columns are just such that reducing the braced length barely works out for axial; adding any moment puts me over on interaction equations. As I change the the brace point to reduce Lb, it only makes the moment worse in the column. If I had a scenario where I didn't need to increase the capacity quite so much, and had beams in both directions, I could easily justify bracing in two directions.

If the footings are oversized enough, could I assume that the base is fixed and use a K=0.8? Or is my only other option to add steel to the section in order to increase the radius of gyration? If so, can I reinforce over the middle 1/3 of the columns? I thought I saw another thread about that earlier. I will try and find that.

Any other suggestions are welcome.

 

[steellion]

-I wouldn't mess with the K-value for an existing column. That would be difficult if the base plate was not designed to take moment. From the AISC Commentary: "For column ends supported by, but not rigidly connected to, a footing or foundation, G is theorhetically infinity but [...] may be taken as 10 for practical designs."

-Honestly, I don't know enough about how a non-prismatic column section would affect the capacity, but at the end of the reinforcement you still have the same axial load, cross-sectional area, K, Lb, and r, so I fail to see how that solves the problem. I would just reinforce the column top-to-bottom and be done with it.

 

[structSU10]

By reinforcing the member you can reduce the K for the whole thing - increasing stiffness in the middle certainly helps improve bucking of the member as a whole. I did a study on this once, and used the finite difference method in Mathcad to determine my effective K for a partially reinforced tube column. I assumed reinforcing extended over the middle 50% and found if my effective EI was 2 times the unreinforced shape, my K = 0.76. I then analyzed the column as unreinforced but with the better K to determine its capacity against buckling.

 

[TehMightyEngineer]

I've reinforced a column using plates to take a jib crane load over the middle third only. Granted the plates were only there for torsion between the jib attachments but strengthened the column overall by increasing the radius of gyration.

Did you consider my note above about filling the columns with concrete? Should be doable and result in a minimum of demolition assuming the foundation can take the extra load.

Maine EIT, Civil/Structural.

 

[steellion]

StructSU10, do you mean that you increased "r" instead of decreased "K"? "K" is based on end conditions; therefore, reinforcing the column would not impact "K" but would impact slenderness ratio by changing "r".

 

[structSU10]

Well I guess I should explain it better. I determined an effective K factor that takes into account the improved stiffness of the member at midspan. Basically the critical buckling load for the reinforced column is higher and can be written in terms of the euler buckling. normally the buckling load is 9.86*(EI/L^2) or pi^2, but when a finite difference approach is used the multiplier could be something like 13.5*(EI/L^2) which can be rearranged to give you K = SQRT(9.86/13.5)=0.855 so the formula for critical load is then pi^2 * (EI/(KL)^2).

Hopefully I explained that well. Attached is a paper on using the finite difference approach to find the buckling load of a non-uniform column.

 

[hokie66]

TME,
I doubt that concrete filling a 4" x 4" column, 19' high, is indeed doable, and doubt that it would help much anyway.