Steel portal frame torsion checks

[moyseh]

Hi all,

Does anyone here check the torsional capacity of their steel rafter in portal frames due to the torsional moment that is normally generated by the eccentricity between the purlin and portal rafter? I’m working to NZS3404

 

[JoshPlumSE]

Not usually. The reason is because there is usually a diaphragm that can (hopefully) restrict the torsional rotation of those members. Therefore, the stick figure modeling of those members is oftentimes not as realistic as we think. At least, that's how I think about it.

That being said, there are times where I add in torsion design for wide flange members. Usually, I do this by hand using the "equivalent tee" analogy.

 

[Greenalleycat]

Where is your torsion resulting from? Uplift or axial forces?

 

[MotorCity]

I'm with Greenalleycat. Trying to picture torsion on a portal frame due to a purlin. Not seeing it. Can you post a sketch?

 

[rowingengineer]

It is possible that the bracing system is such that your bracing lines are at the purlin level and not mid rafter resulting in a torsion from eccentric loading. When I use the detail of bracing at purlin level, I try take the eccentric torsion load out with smart detailing with fly bracing or similar, and also check my rafter for the additional torsion, generally it isn't a huge force and because in eq the rafter isn't working that hard.

 

[Greenalleycat]

Kinda shitty sketch here, but I'm assuming this is the scenario present
I have checked this connection in detail once, in the context of a precast industrial building

Precast walls all round with concrete end walls and portal frames as the transverse lateral system
Roof bracing in the end bays (or first bay back) is used to take longitudinal end wall loads to the in-plane concrete shear walls
In this system, end wall loads are usually generated into a specific strut (a 150UC in my case), and roof axial loads etc are accumulated through the cladding and come out into the steel purlins at the points where roof bracing nodes in

In this configuration, there is commonly a vertical eccentricity between the axial loads generated in the 150UC, the roof bracing (bolted to either the 150UC or transverse portal frame flange) and axial loads accumulated into the purlins
I analysed this connection based off the torsional stiffness of the portal (not much), and the relative major axis bending stiffnesses of the purlins and 150UC strut
I found approx 80% load went to the strut and 20% went to the purlins based off this distribution
This then resolves through the bolts... which meant I had to size the bolts up and increase their vertical spacing to absolute maximums to get the moment-induced shear to resolve

The portal torsional stiffness is so low that it basically does nothing to contribute to this resistance, particularly if you bother to model in every purlin (typically 1.2m spacings here)

Frankly, at the end of the exercise I didn't believe the results I'd come up with
If the resultant torsion really does follow this load path then struts and purlins should be falling out of the sky in earthquakes as the bolts shear out trying to resolve the torsional moments

Probably I missed some secondary load path somewhere

 

[MotorCity]

There was no mention of lateral loads so I assumed we were only talking about gravity loads.

However, the eccentricity shown in purple does not exist. Any axial load in that purlin would simply get transferred into the next purlin. The beam would not see that axial load. I do agree with the torsion created by the lower beam strut framing into the beam if thats what we're talking about.

 

[Greenalleycat]

The eccentricity does exist if that is the node where the roof bracing connects in, or else that load cannot make it out to the in-plane shear walls (barring funky roof cladding situations)
I have not drawn in roof bracing as the sketch was getting complicated enough as it is, but I have noted that the roof bracing connects to the 150UC in my sketch

Edit: also, 3404 is the NZ steel code (where I operate) and all of NZ is a seismic zone to some extent, so my thinking defaults to lateral loads in such situations as the scenario I drew is very common here

 

[KootK]

Based on what you've described, it sounds as though there is indeed the eccentricity. More over, it sounds as though the eccentricity might actually be from the centroid of the purlins to the centroid of the 150 UC. At the same time, perhaps the 150 UC's can be thought to provide torsional bracing to the portal rafter such that the impact of the eccentricity will be small?

 

[Greenalleycat]

@Koot, yep, agreed. The eccentricity varies in each direction though, if the compression force has to go down into the 150UC to get to the bracing then it's a larger eccentricity than if it only has to go roof bracing at top flange level
The 150UC does torsionally brace the portal (especially when full depth cleats are used inside the portal flanges)

The issue then becomes following that eccentricity out
That was the (largely academic, IMO) exercise I followed through
Effectively, that whole UC-portal-purlin connection wants to rotate in space but is held there by the major axis bending induced in the UC and purlin (and old mate, the roof cladding, of course)
This moment then has to resolve into a moment on the bolt group of the connections at the UC and purlin ends
This resolves into vary large shears (in addition to those from gravity and EQ axial loads) that theoretically write off the capacity of typical bolting configurations

 

[KootK]

When you speak of the 150 UC and the roof bracing, those members are part of the same horizontal truss, even though they are not strictly co-planar, correct?

Does your building have roof level trussing at both ends of the building? I believe that one of the benefits of such schemes may well be to reduce the impact of the kind of thing that we're discussing here.

How often are the 150 UC spaced?

 

[Greenalleycat]

You are correct.
These 'truss setups' are usually in the end bays only (i.e. two per building) for smaller industrial setups

If the roof bracing connects at the UC height then the eccentric loads are massively reduced as the end concrete wall loads do not have to travel at purlin height at all
Typical setups would have the roof bracing connected at the top flange height of the 150UC strut, so there is technically 150/2 = 75mm of eccentricity
In this case the eccentricity remains within the 150UC section, which is quite robust and should easily be able to resolve the moment out into small vertical forces at each end
Conventionally, this is ignored in design

This was a building I was assessing rather than designing, so in this case the roof bracing actually connected to the portal top flange, which was therefore half a portal section depth higher than the 150UC centreline (= 310/2 = 155mm)
This introduces extra eccentricity which has to transfer via the bolt groups of the purlin and 150UC in order to resolve out
I do not think that this configuration should be ignored as this is when the large moment-resultant shears can occur in the bolts

 

[KootK]

I do not think that this configuration should be ignored as this is when the large moment-resultant shears can occur in the bolts.

I'm inclined to agree. Thanks for sharing your observations on this. I'm sure that most of us have been guilty of hand waving away eccentricities in the context of horizontal trussing at times in the interest of attempting to be "practical". Clearly, there are limits to that.

 

[Greenalleycat]

Gotta admit, it was one of those things that just popped into my head one day that made me go..."crap now I gotta put a number to it"
In practice, it doesn't seem to have been an issue in the industrial buildings here (in Chch, NZ, there are lots of industrial buildings that got a hell of a shake in 2010/11)
The running industry theory is that the roof is the MVP in resolving these things

Edit: sorry, Moyseh, this discussion has hijacked your thread. I hope that what I posted was indeed what you were asking for input on...

 

[moyseh]

Thanks for the comments. All helpful stuff. I’ve encountered often buildings that have roof bracing fixed to the bottom flange of the portal hence introducing more torsion however I’ve never put a number to the torsional capacity of the rafter. Do you guys typically calculate that with a software/by hand or your own spreadsheets?

 

[Greenalleycat]

Do it by hand generally. In the case of this thing I investigated the torsional route of the portal by modelling it in SpaceGass. That was largely academic though, the analysis of the bolt group etc I just did with hand calcs