Bracing connections with hollow sections

[PXC]

I wonder if anyone on this list has addressed the design of bolted connections in which a plate is slotted into the hollow section, and this is bolted alongside the support plate that is fixed to the beam or column, the bolts being in single shear. The design of these is covered in AISC (Australian) design of structural connections. I would be interested in discussing some aspects of the design model that I think are wrong.

 

[dbuzz]

I have previously designed slotted CHS bracing connections to the AISC guide.

What are your concerns with the design model? I'd be interested to hear them.

 

[PXC]

Refer Design of Structural Connections 4th edition: Sections 4.11 and 5.11

1. The design model discusses the eccentricity e but does not include it. I think that the eccentricity is in fact significant.

2. Lb is taken as the distance from the support plate to the first row of bolts. For unstiffened plates slotted into hollow sections as per the figs, I think it is appropriate to use the length right up to the end of the hollow section.

3. ke is taken as 0.7. I think it should be 1.2 ie as for a sway member.

If my comments are correct, then the AISC design model seriously underestimates the capacity of these connections.

 

[dbuzz]

ECCENTRICITY
Which eccentricity do you mean:
- In-plane eccentricity due to offset of member centreline from bolt group, weld group and gusset plate centrelines
- Out-of-plane eccentricity due to misalignment of cleat and gusset plates

For CHS connections:
- In-plane, the member and cleat centrelines are coincident with bolt group, weld group and gusset plate centrelines, i.e. no in-plane eccentricity
- Out-of-plane, the eccentricity is only due to the cleat and gusset plates being mounted “side-to-side”. Most references agree that this is negligible for static strength and AISC's LRFD Manual of Steel Construction states that this is supported by testwork on similar statically-loaded welded connections. You can avoid eccentric effects by adopting a “tongue-and-cheek plates” connection, which keeps the cleat and gusset plates coincident. In my experience (mining, minerals processing and industrial structures) this is typical for CHS connections anyway.

EFFECTIVE LENGTH FACTOR
The design model, which uses ke = 0.7, is based on AISC’s Engineering for Steel Construction and Thornton’s Bracing Connections for Heavy Construction, which use an even lower effective length factor of 0.65. Your assessment that it might be treated as a fixed-fixed sway member instead, with ke = 1.2, might have some merit. I’d have to read the references before making a comment though.

 

[PXC]

1. e is what you refer to as out-of-plane eccentricity. eg if the support plate is 20 thick and the plate slotted into the hollow section is 20, then e=20. I have made some simple models to detemine the bending moments caused by this eccentricity and then calculated the stress and found they are significant.

3. I saw another thread a few months back that was referring to similar designs to AISC (USA). I got hold of the reference and it used (I am relying on my memory now) ke = 0.65 when the support plate was fixed along 2 orthogonal lines eg. the top of a beam and the column it is connected to. When it was connected along only 1 line such as the top of a beam, then it used 1.2. The AISC (Aust) design model uses 0.7 for a support plate fixed along 1 line.

A value of 1.2 is consistent with Fig 4.6.3.3, AS 4100 sway member, where the "member" is the plates between the support and the end of the brace.

2. This same USA reference showed something I have seen in a number of pictures from USA. They show the brace as an I-section with out-of-plane stiffening (eg. part of the I-section or angles attached to it) extending from the brace along the length of the bolts. In this case, it is appropriate to take Lb as the distance from the supporting member (eg top flange of beam) to the first row of bolts, measured along the centreline. The AISC (Aust) model clearly shows a plate that is not stiffened.

 

[PXC]

dbuzz, I found this in the readme file for Limcon V3, dated August 2003. It seems to support my comments above. My comments are in brackets:

Bracing Cleat Connection
------------------------------------
From 07/08/2003, Limcon V3.02 calculates the compression capacity of cleats more conservatively than in previous releases (i.e. previous releases were unsafe).

In computing the compression capacity of the bracing cleat connection determined by out-of-plane buckling of the plate or connected plates, Limcon departs substantially from Ref. 4 and Ref. 15. The effective length factor, k, recommended by Ref. 4 and Ref. 15, are 0.7 and 0.5 respectively. These values can only be justified when there is restraint against out-of-plane buckling of the connection. The corresponding compression capacities, if shown, are informative only.

For AISC LRFD the k factor is 1.2 in accordance with Ref. 16, while for BS 5950 and AS 4100 it is 1.0. For eccentrically connected plates the compression capacity is determined according to the configured design code, taking compression and bending interaction into account. The properties of the equivalent column are determined as the minimum of the corresponding values for the two connected plates.

NOTE: In checking compression capacity the small eccentricity may be ignored where a member is connected directly to a gusset (eg web of a UC to a gusset) but it must not be ignored where gusset plates are overlapped. (ie slotted hollow steel section)

References:
4. Design of Structural Connections - 4th Edition
- T.J. Hogan & I.R. Thomas - ASI - 1994.
15. Design of Structural Steel Hollow Section Connections
- Volume 1: Design Models - 1st Edition
- A.A. Syam & B.G. Chapman - ASI -1996.
16. Hollow Structural Sections Connections Manual
- American Institute of Steel Construction - 1997.

 

[dbuzz]

PXC,

Thanks for closing the loop. Sorry I've gone quiet on you, it's a particulalry busy time at work at the moment.

This is certainly something for designers to be aware of. I know I definitely will be in future. Having said that, for even quite small CHS braces we normally use the "tounge-and-cheek plates" connection.

Good on you for following it through.

Cheers,

dbuzz

 

[engsys]

I am involved in the development of the program, Limcon V3, mentioned above, and I was interested to see these posts on slotted plate hollow section connections. I have raised similar concerns with the Australian Steel Institute, publisher of the two design manuals referred to. The situation is covered in "The Problem with Lapped Gusset Plates" at:

http://www.microstran.com.au/lmc_brac_cleat_comp2.htm

.

I believe there is an important difference between a gusset plate connected to a stiff member (angle or channel, say) and a slotted plate connected to another plate (gusset). In the former case, nearly all the eccentricity moment is resisted by the stiff member - in the latter case, the moment is shared about equally between the two plates. In this situation, plastic hinges develop in the plates at a compression of about 40% of the yield capacity of the plates.

As for k factors, in my opinion, k should be 1.2, as is common practice in the US and Canada. k = 1.0 is used in UK and the rest of Europe, probably in recognition of the existence of some restraint from other members at the connection.

I don't think unstiffened plate on plate compression connections should be allowed at all. Follow the links at the above URL to see the kind of concentric HSS bracing connections commonly used in Europe.

 

[PXC]

Thanks for that engsys
It was interesting to see the photoes of concentric conections in Europe. I have never seen them in use and have always been reluctant to use them because of durability, fabrication and erection issues.

 

[engsys]

Expensive is better than collapsing!

 

[dbuzz]

I could see the quality of surface treatment in the "slot" being a problem for minerals processing, oil & gas and marine applications.

By the way, it's good to see some Australian discussion on this site!

Cheers,

dbuzz