Partial Snow Loads

[aaronmcdevitt]

Hi all, I have been trying to find out a good way to design a steel roof system in RAM for snow loads for some time.

Problem:
RAM treats snow loads as live loads, and creates a "worst case" scenario by "skip loading" - applying all or nothing on cantilevers/backspans. However, according to ASCE 7, balanced snow loading only needs to be staggered 50%/100% (the section on partial loading for multiple spans). This results in RAM being overly conservative (by 20% in a simple test structure).

Solution attempt:
I am assigned to try to come up with a reasonable workaround, but so far I don't think there is a reasonable solution. Does anyone know whether any of the live loads (storage, partitions, etc) are NOT staggered? If so, I can apply this load as a uniform load equal to one half the snow load, effectively causing RAM to stagger my snow loads 50%/100%. I cannot use a dead load for this due to deflection criteria.

I hope RAM isn't completely worthless for snow loads...

- Aaron

 

[SethGuthrie]

This is a fairly common problem for snow loads in cantilever-suspended span roof framing systems. There is not a good solution since all of the live load types are patterned as you noted. I do sometimes recommend taking 50% of the live load and adding it to the dead load (which is not patterned), but this has some drawbacks like different load factors to contend with or deflection issues.

If you make all of the framing lateral and process the designs in Ram Frame, you have more control over the load combinations, but you still won't get deflection checks going that route.

Other than the dead load trick you might have to do some manual checks by superimposing the moment demands from the live load patterns.

 

[SteveGregory]

Aaron, Being 20% on the conservative should not make any real difference in the cost of your building unless your balanced roof snow load goes over 100 psf. If the roof snow is more like 30 psf, then 20% more makes it 36 psf. That sounds like a good thing to me.

Deflection and ponding coupled with clogged roof drains may be a more important design criteria to consider.

Steve

 

[aaronmcdevitt]

I have nine grids about 220 feet long. That's about 2000 feet of girders. Comparing strength sizing for Lb= 6 ft (our joist spacing) and Mu=600 k-ft vs. 720 k-ft (20% higher), I get W21x68 (or 24x68) vs W24x76. 8 lbs times 2000 ft = a possible 16,000 lbs extra weight in girders.

In reality it isn't this high. A few girders are deflection controlled, and a few are controlled by minimum flange width. Still, I'd like to save a few tons of steel if possible. Anyway, my boss wants to figure out if RAM is worth using in the future, and to compare RAM sizes with RISA sizes. We would rather not specify 10% more steel than already required - after all, the snow load is a rare event, there is a huge load factor, material strengths are very conservative (available bending strength doesn't account for Ry or strain hardening), our dead loads are conservative, etc. Why add in extra conservatism?

All in all, it seems this dead load workaround is good in this case, but not in cases where live load deflection might govern.