Help me understand this roof framing

[SinStrucEng]

We have a rafter framed roof with a ridge plate (rather than ridge beam). There are few collar ties, but the rafter ties (which act as the ceiling joists) span in the direction opposite of what I would expect (they are perpendicular to the collar ties and parallel with ridge). There is a knee wall at around 1/2 span of the rafters, which simply sits on the ceiling joists below.

What is stopping the exterior walls from splaying out? Exterior facade is brick and in great shape. Structure is from the 1970s. We were hired to design a new staircase/opening into the attic so that the space can be used for storage and whatnot.

My intention is to cut the existing ceiling joists back a touch (36" or so) from the CMU firewall to fit a staircase from the second floor and to simultaneously upgrade the joists (either sister or replace with new of a different size) to make them adequate to support the expected dead/live loads. Current ceiling joists are only 2x6 at 16" so they will become either 2-2x6 at the least or maybe 2x8, 2x10, etc (this analysis isn't complete yet).

My main worry originally was that by cutting or removing the ceiling joists, you'd compromise the roof. But it appears that these joists are literally only supporting the ceiling and insulation, and don't tie the roof together at all... They're even lapped over interior partition walls. What am I missing? Is this roof magic? We live in a heavy snow area too.

 

[L-H]

How does gypsum board on the rafters help? Adding those gypsum boards will actually add more dead load to the system and hence more thrust.

 

[lexpatrie]

No research data on that I've ever seen, it's an engineering hardware. I.e. it's assumed in the absence of testing or research to work that way. This mentality shows up almost constantly in this forum trying to justify how these dodgy roof framing systems work when the math and engineering for the established load path does not satisfy.

While I am sympathetic as I've done these kinds of analysis and the rafters fail by say 300% per code, best case, even ignoring deflection requirements, if you presume the rafter is somehow braced at the ibtermediate bearing point despite a lack of blocking (one could potentially add it, after all)...

Something is assisting these roofs that's not accounted for in traditional analysis, or they've never seen the design loads, the alternative load path isn't proven or tested. It's assumed. You can draw all the free body diagrams you like, but it's a leap of faith, not engineering.

 

[XR250]

Also, if you are sheetrocking the rafters, they need to be L/240 now. So another way you are touching it.

 

[lexpatrie]

That I think would be for the transient load alone, snow or wind, for D+L it would be L/180.

This is more likely IRC, which is more vague about what loads apply to the deflection limits. At least I got the feeling it was residential.

This is going to be quite challenging to get the roof assembly to work, in an engineering sense, if you ask me.

 

[XR250]

Ah, I guess the NC version is different.

Either way, if that sheetrock starts cracking, there is going to be some finger pointing.

 

[lexpatrie]

I was tired, so I only posted the IBC language.

Your IRC is probably similar, the language on the situations doesn't match between IBC and IRC ("nonplaster" versus "including gypsum board" for example) and the IRC doesn't clarify by adding a load case.

Source: 2021 IRC, Table R301.7

Darth engineer can apply it however they feel (most favorable) and Obi-Wan engineer can apply it in the most restrictive sense (least favorable).

You may also be saved by the sequencing of the ceiling finish as it may have predeflected a bit for the roofing and whatnot before the ceiling is installed.