Effective length of beam subject to lateral torsional buckling 1

[valleyboy]

Hello all

A quick poll to gauge oppinion :-

A pair of parallel simply supported beams carrying the inner and outer leaf of masonry construction in a domestic 'knock-through'. Inner beam carries one storey of masonry, roof of main property, and first floor joists. Outer beam carries mass of existing outer skin and roof of new single storey extension.

The members are UB's (or I sections for the American audience) and are battened together with mild steel plates to the top and bottom flange at mid-span. The top flange level of the beams is below the existing first floor construction and as such the flanges are not connected to any horizontal diaphragm in the structure. The members are supported off masonry piers at their bearings.

What is everyones view as to the effective length for lateral torsional buckling?

(I had this discussion with some colleagues in the office today and was surprised by the range of oppinion !!)

Regards

VB

 

[SlideRuleEra]

IMHO - If the first floor joists are more or less perpendicular to and connected to the inner beam, then the unbraced length of the inner beam is equal to the spacing of the floor joists (perhaps 16", or so).

Then, the unbraced length of the outer beam is half it's span since it is properly connected to the (well braced) inner beam at midspan.

http://www.SlideRuleEra.net

 

[valleyboy]

Thanks for your post SRE.

Just to clarify - the first floor joists are not connected to the inner beam, and the top flange of the steel beam is set at a lower level than the adjacent floor structure.

Regards

VB

 

[CowboyBill74]

valleyboy

I am having a hard time visualizing this type of construction, but if i do have the correct scenario then the only material directly above either beam is masonry. If that is the case then I normally do not consider masonry as a brace for the top flange and therefore would consider both beams to be unbraced over their entire length.

However if the first floor diaphragm is close to the top flange of the beam (say within one course)and the floor joists are attached to the masonry then one might be able to make the arguement such as SRE has outlined, but that is making the assumption that the masonry is securely attached to the top flange of the beam.

Interesting question I will keep in touch with this thread to see what others think

 

[xxpegasus11]

Not surprised you have so many opinions - Easier to take a practical approach than academic. In a situation like this and I come across it frequently, I never take the floor as providing lateral bracing even when the joists are at beam level, or even fixed somehow to the beam which would usually be with joist hangers because a number of other factors come into it - basically I cant rely on fixity of joists at other end, flooring not being altered etc With beams resting freely on bearings you also only have partial torsional restraint at ends.
Your battening with plates - presumably to offer some deflection compatibility - what is the design of this? is it welded on site? bolted?

 

[miecz]

Seems as though the batten plates would restrain both beams from twisting. I'd check each individual section with an unsupported length of L/2. Then I'd check the composite section, assuming both beams will twist as a unit, with an unsupported length of L.

 

[pba]

Vallyboy,

I guess you work in the UK? (I also guess that you are Welsh!). I assume that you are designing to BS5950.

To assume a lateral torsional restraint to the beams you need to comply with the conditions given in clause 4.3.2.1 - Thus you need sufficient stiffness to inhibit lateral (i.e. horizontal) movement. This is a debateable point (hence your posting). There are two possible ways that this stiffness can be provided in your example - From the adjacent roof or floor joists or from the masonry wall itself. Normally it is considered a bad idea to rely on timber to restrain steelwork (At least in the UK). The wall itself is more promising. In order to determine if the wall can indeed be considered as a restraint then you need to look at clause 4.3.2.2.1 and determine the value of 2.5% of the compression force in the top flange. This force is the udl force on the edge of the masonry. You would need to determine the wall strength to ensure it was capable of carrying that load in addition to the other forces present on the wall. My experience is that determining masonry strengh as adequate in domestic construction is a near impossible task as the designs are based on Building Regulations "deemed to satisfy" clauses rather than design codes. You thus have no other alternative than to assume that the wall does not restrain the beam.

Now lets look at table 13 to determine the effective length. I assume the beams are sat on concrete padstones and possbly bolted down but that the end of the beam is otherwise left as rolled. I'd say that the compression flange was unrestrained at the support as there is no difference between the support and the mid span. For this to be the case it does not matter if the flanges are free to rotate on plan or not as you are already in the bottom half of the table. As I read the code you now have two considerations. If the beam is bolted down it is the upper box, if it simply rests on the padstone than it is the lower box. Here I take issue with the code as this is a beam resisting a udl load from a wall over and this will serve to anchor the beam in position so I would argue the top box regardless.

The final thing is destabilising load. The load IS on the top flange. However, if the beam moved sideways it is unlikely that the wall would also move as the wall is relatively rigid. On this basis I'd say the load was not destabilising. I usually adopt a slightly different detail to you in that I try to use channels rather than UB's as by placing these in a back to back configuration I get equal and opposite torsional effects which lends further weight to the arguement of a non-destabilising load.

Finally - you are lacing the two beams together at mid span. This does not provide a point of restraint as the lacing only transfers force to an equally unrestrained beam.

Like you I have heard all opinions on this problem but remain unconvinced - This is one of the few cases where I don't think any of the "reduction" factors are valid.

 

[patswfc]

valleyboy

I agree with most of pba's comments. The only thing which i would question is whether or not the load should be treated as destabilising.

I would either design both beams seperatly with an effective length of 1.0LLT + 2D(normal) or 1.2LLT + 2D(destabilising), depending on how you veiw the action of the load (this will give very conservative design, i've also assumed that the beams are bolted to padstones), or provided that the connection between the UB's and plates are adequate, then you basically have a combined section, which is closed. Thus ltb isn't an issue.

What has been the general concensus in your office?

As a matter of intrest how long are the beams in question,
also make sure that you use appropriate deflection criteria.

 

[valleyboy]

The oppinions offered in this forum have been pretty much the differing oppinions in our office.

(And in answer to pba's queries, yes I'm in the UK, Welsh, and designing to BS5950!)

I was surprised with the range of oppinions, because I thought this was a bit of a 'no-brainer' - the battens offer no lateral restraint as they simply tie the beams together making them mutually dependant, and offer no resistance to lateral movement of the top flange relative to the supports as required in the code (in fact the code states that tying two un-restrained beams together is not adequate!)

I have also come across the argument offered by pba that frictional resistance between the flange and the wall offers the necessary restraint. I am a little uneasy taking restraint from the load in this way. In addition, the SCI Publication 'Steel beams and columns - common cases of restraint' suggests that in such arrangements no restraint is offered.

As such I think that the effective length of the beam should be 1.0L.

I note pba's and patswfc's comment regarding de-stabilising loads. I believe this exists if the beam and load are free to rotate should the beam buckle.

I am relatively happy that the upper storey of masonry is adequately restrained by the first floor construction and trussed rafter roof over, and as such would not be able to 'rotate' should the beam start to fail.

Finally, I would be happy if the beam were mechanically fixed to one of the floor joists at mid span to reduce the effective length to L/2, as I believe that a conventional floor will be adequate to act as a diaphragm to resist 2.5% of the force in the compression flange.

Interesting one though !!

VB

 

[xxpegasus11]

I agree with most points made.
I dont feel the joist offers enough lateral restraint to justify me reducing effective length so I ignore it.
As pba says, this takes us to Table 13 with two choices of 1.0L+2D or 1.2L+2D and not 1.0L as valleyboy. I then also feel like pba that 1.0L+2D applies. I agree that the load in this situation is not destabilising.

Bolting to padstones - Do you usually bolt to padstones? if so, what is benefit and how does it translate into a design parameter? As far as "destabilising" I am happy that the masonry wall load is sufficient so I dont need bolting for that. Bolting which provides some kind of fixity leads me to a moment transfer which I would rather avoid. I have seen restraining the beam from rotational or longitudinal movement ( including expansion) leading to cracking induced at wall support, so in a case like this I omit bolting altogether. Is this your practice?

Batten lacing - provided or not provided? The way I look at it is :
effective length - no reduction as we all agree that it doesnt qualify
composite action - beams are not designed as composite unless one made them so through providing adequate load transfer. I assume we are talking about a small plate, say at 1 or 2 places ? am I correct? If so, I wouldnt design it as composite.
practical installation - if prewelded together that makes it very difficult to install so it must be done afterwards. I then assume that its not welded on site but bolted? Is that the case? Also that the connection is just rule of thumb rather than based on calculation? More often than not I have therefore omitted any lacing.

 

[patswfc]

I just read the initial post again. The beams are laced together at mid-span. Didnt take this in first time. I was wondering why people were saying the effective length could be taken as l/2.

Thus my comment about designing as a combined section is rubbish.

Therefore design beams with an effective length of 1.0LLT + 2D.

As you say valleyboy it is a bit of a no brainer.

As for using the floor joists to restrain the beam, this is generally avoided in the UK as pba has previously stated. I would avoid using the joists as restraints for the reasons that xxpegasus11 has given.
Having said that SCI publication "P093 Lateral Stability of Steel Beams and Columns Common cases of restraint" does give an example showing timber floor joists restraining a beam, but I wouldnt go there.

 

[pba]

There is no technical reason why timber cannot be used to restrain steelwork. The reasons why it is not common practice may be largely historic. In the UK we don't tend to use sheathing as is used in the US and otherplaces so roofs generally do not have the required 3 dimensional rigidity. Timber floor boards are now most commonly replaced with chipboard which could, if suitably designed, be used as a diaphragm. You still have the problem of ignorance on the part of owners and small building firms who are unlikely to understand the importance of such elements of the building.

Home owners typically feel happy to mess about with wood but are fortunately reluctant to do the same to masonry and steel.

When I graduated I worked alongside an engineer heavily involved in structural surveys. He had a library of photographs of building defects. Most of which were caused by stupid people - home owners and builders - not understanding what they were doing.

 

[pappyirl]

I agree with the consensus arrived above.

xxpegasus11 - if you don't bolt the beam to the padstone, what do you do with it?

 

[xxpegasus11]

pappyirl - just rests on padstone - I dont know what span we are talking about in the case here but assuming a normal domestic opening this is no more than just a longer lintel.

 

[apsix]

It's worth reading 'Lateral-torsional buckling of steel beams in domestic buildings' by Harris, published as a Technical Note in The Structural Engineer Vol 78/No 7, 4 April 2000. Link is;

http://www.istructe.org/thestructuralengineer/hc/Abstract.asp?PID=6473