Lateral Restraint of Roof Purlins

[jeg]

Consider a house roof designed to maximise useable space - no trusses.
The concrete tiles rest on battens which rest on timber rafters.
The rafters span from eaves to apex and are supported by steel PURLINS at mid span. The rafters are bird mouthed over the purlin and "may" be nailed to a timber runner bolted to the top flange of the purlin.
The purlins are supported on the top course of blockwork walls rising from the floor below. The amount of building-in is negligible.

Question: reference clauses 4.2.2 and 4.3.3 in BS5950 can we consider the purlins as having "full lateral and torsional restraint" ?

I consider them to be unrestrained and to have an effective length = 1.2L + 2D Is this too conservative ?

I see many instances where the effective length has been taken as L. These structures have not experienced any problems that I am aware of. Is this just good luck?

 

[SteveNapper]

Jeg,

I'm a Tech Support Engineer at Ward Building Components in the UK. If you visit our website,

http://www.wards.co.uk,

you will be able to download a copy of our 'Multibeam' handbook. This has, among many others, details of our standard cleat / purlin connections. This may help.

Also we tend to use an effective length of 0.85L, providing that the connections are 'substantial'. This is from Table 9, in Section 6 of BS5950 - 5 1998. Then again, our load tables are based on tested values, not calculated, as our competitors.

Hope that this helped.

Steve

 

[RMalaska]

The lateral and torsional restraint is more dependant on the type and nailing of any sheathing (plywood) on top of the rafters. I think it may be overly conservative to be using an effective length of any more than 1.0 for rafters with sheathing nailed at 12" oc. I am not familiar with BS5950, but basic engineering principles are the same the world over.

 

[jeg]

Thanks for the info, RMalaska.
In the UK it is not usual to have sheathing nailed to the rafters.
We drape waterproof membrane over the rafters and hold it in place with battens nailed to the rafters. The tiles are then hooked or nailed to the battens. This is not nearly so restrained as your roofing system.
Hence the reason for my query.

 

[ishvaaag]

And also an interesting struggle between what the codes may demand and what providers may recommend in order to make their products competitive. If all goes well (usually within proprietary systems) one usally may use what the fabricator says, because only unknowingly they would recommend something out of the code, which is rare.

In any case, if you can model your construction with reasonable accuracy and include in the analysis end node PDelta effects, you can use K=1 for sure. Less than that you may based on some technical ascertainment, such recommendations in books, or preferably statements in applicable code.

If you in more include some intelligent in-member imperfection, account in some way for material anelasticty and divide your member in a moderate number of submembers, you might end even in being able to do away even with other checks than pure strength of materials. That is, you don't need at all a (bending) K anymore. But of course the codes don't use to be explicitly supportive of this, and of course, as any way of proceeding, can be criticable on some accounts.