Wood open roof 2

[milkshakelake]

Client wants an open gable roof. I don't like to do it generally because of lateral thrust; I'm more comfortable with having horizontal attic joists. Are there any references for how to design something like this? I was considering having some horizontal timber beams at like 8'-0" on center to deal with the thrust, beyond the window so the structure wouldn't obscure the facade.

 

[XR250]

The structural ridge is not going to alleviate the thrust. The deflection of the ridge beam will be resisted by the axial stiffness of the rafters.

I agree the way you suggest is the most common way, but it is wrong IMO because it doesn't eliminate the thrust, thus the 5 leg analogy.

The ridge is supported vertically by a bent flitch or I-beam at the gable end - thus eliminating thrust. Similar to this...

https://www.finehomebuilding.com/project-guides/framing/open-up-a-ceiling-with-a-steel-sandwich

 

[SWComposites]

Guess I don’t like all of the flat valleys that are prone to leaks, particularly in snow country.

 

[Enable]

MSL: visualize an I-beam with baseplates on each end being heated and bent to the shape of a perfect arch like you'd see behind newly weds in their staged photographs. Now place that same steel shape with the baseplates firmly in contact with, but not anchored to concrete. Then place a light load on top of the of the arch. What happens? Basically nothing. The "thrust" as it were is resolved by the steel member internally since it has moment continuity throughout and the bases cannot kick out due to the moment fixity at the load location. Now add a really big load. What happens? Probably things start to move since we've gone beyond the moment capacity of the cranked I-beam (and it starts to split but with no positive connection to the concrete lateral displacement results). But up until that point, we experience no or limited movement.

In the case of a pinned connection at the ridge (as is typical with wood to wood connections) you'd get thrust, but not the way XR250 / Drift are suggesting. At least this is how I interpret their advice.

 

[milkshakelake]

@SE2607 That's what I meant. Maybe worded imprecisely. I figured the joists are restrained by the diaphragm, and there's never not a diaphragm for roof joists. I just reworded what you said in a way I understand. I agree that if there's any type of connection between the two diaphragms, it should be okay. I'll have to figure out how that would look like in terms of wood runners and getting the angles to work.

@XR250 Thanks for the link! Can you humor me about how the free body diagram will work out? The way I'm visualizing it, because it's stiffened with a flitch plate, it's akin to a horizontal beam. We're not worried about it opening up at the top. So it serves the purpose that the horizontal member of a triangular truss. At the same time though, there is an axial force. It can't just be bending, because it's inclined. So I'm getting stuck at the point where an inclined member meets the column, unless I think of it as a horizontal beam. (Sorry if there's something obviously basic that I'm missing.) I'm assuming pin on one side and roller on other.

@SWComposites If I were rich, I'd have a gigantic open gable roof. I saw a house like that and it's beautiful, but it was done with steel. For most people, I imagine that the cost would be too high.

 

[milkshakelake]

@Enable That makes sense, thanks for the explanation. I'll have to do some thinking about where the forces go; just having trouble with the horizontal element. Intuitively, I understand what you mean.

 

[lexpatrie]

Get them to split the window at the center and put in a post and a ridge beam, these other "schemes" are a fantasy.

 

[driftLimiter]

Perhaps Lexpatrie has a point. A thin post that is like a window mullion is the guaranteed way to eliminate the thrust.

Your point about the cranked beam is a valid one. The axial stiffness of the legs of the beam will for sure take some load.

Another idea would be to use the cranked beam, then get a horizontal (mullion?) In the window somewhere near the top to reduce the cantilever of the jamb columns. Maybe a horizontal obstruction to the window above head height would be more acceptable than the vertical mullion Lexpatrie is suggesting ?

At the end of the day this is not a problem of engineering as much as architecture. The owner/arch wants a big huge window, they have to pay for it in framing.

The cranked beam horizontal thrust problem however is one of compatibility. If the beam is pin-roller it will have some horizontal displacement. That displacement is compatible with the cantilever column. As the cranked beam becomes more and more flexuraly stiff, the free displacement horizontaly becomes less and less.

 

[lexpatrie]

Since I have a project where I'm trying to fix a dead valley I might as well point that out as well. You want overframing there to force drainage to the far corner of the eave. Or rather, point that out in an email to the architect and file the email in your correspondence folder.

One would like to think eventually the "promise" of fully 3D rendered projects will somehow alert people to bad idea framing, but we are still waiting for that to happen, aren't we.

I've gone through convolutions very similar to this, including RISA 2D models (and all the debugging that comes with that, rough sawn, new construction, Lb etc, etc, etc, ) fairly recently. It seems like you can get it to work, and then the reality of (3) 2x10s @ 16" o.c. sets in for rafters.

The framing shown is a large room, so it isn't like you can cantilever a ridge beam with a pair of columns anywhere normal, and the other approaches are going to produce a lot of deflection and forces you probably can't transfer in (normal) wood framing. The wall won't have the stiffness you want it to have to limit the spreading, and it will likely overstress the studs in bending by a significant amount. Then there are the forces at the double top plate you are trying to transfer (laterally) and the uplift that's already there.

If you go with something that takes fourteen details to explain, who's going to inspect it?

Side note: That XR250 article looks pretty interesting.

 

[Flotsam7018]

MSL - There is shear in addition to the axial, and the horizontal components cancel. The horizontal movement at the supports does have to be accounted for with this model.

What SE says is true, as is the case with any ridge beam - the horizontal movement has to be there in order to engage the beam. I'd be comfortable with that assumption here and moving on. It could certainly be detailed to transfer the thrust into an adjacent diaphragm, but that seems overly complicated IMO, though in reality is very likely to be what happens.

 

[Enable]

MSL: I'm with you. My explanation makes sense (to me) from a qualitative perspective but quantitatively, I would be pressed to put numbers to it in the immediate since I haven't really thought too hard about something like this before. If I had to do something I think I would say that from a global perspective, if we have moment continuity at the ridge and we don't exceed member capacity, the center of the arch cannot move laterally nor vertically (ish since it obviously will). I would then take half the arch and assume the base as a roller and the middle fixed (though we use the axial force from a pinned-roller model). Thus the horizontal component is resolved via flexural stiffness of the arch leg itself. The more flexible it is, the greater amount of deflection.

I might then treat it as a straight member for ease of deflection calculation (effectively a cantilever - you could remove the roller for this) and would back-calculate required stiffness based on the allowable wall movement. For a steel arch and loading from standard light-frame construction I imagine the deflection could be quite small for reasonably sized members (hence why XR250 suggested even flitch beams might work). I'd be curious what an FEM model shows for any of our members who are so-inclined. I can offer thanks and beer if you ever come to Ontario!

 

[milkshakelake]

@lexpatrie No more big window = dead milkshakelake. Anyway, I don't think it'll take 14 details to explain. I think I can easily build in some redundancy by engaging the shear walls, cantilever column action, and the adjacent roof diaphragms, which is like 3-4 extra details. I don't think it's super complicated construction-wise. It's just making my brain do some backflips thinking about how to calc it out. It's something I feel in my gut will work, but will need to think about how to justify. I might do something like 50% of the thrust goes into the columns and shear walls and 50% goes into adjacent diaphragms. Wood is a bit wishy-washy for me anyway, with its weird load paths. The good thing is that this particular township isn't very strict with calculations, unlike where I normally practice, so I have leeway.

@driftLimiter I don't think a horizontal member will work. Client wants a big window. But anyway, I'll present the option as a cost-cutting option.

@Flotsam7018 Although connecting to an adjacent diaphragm is tricky, the horizontal movement thing is very tricky to calculate as well (at least for me), because you need to engage the shear walls and/or the column. I think I'm going to skip a few steps, do some quick calculations, and just connect everything nicely and let the redundant connections do the work.

@Enable Wow...the thing you described is like a full day of work for a barely qualified engineer like me. I get what you mean, and that would be somewhat close to an accurate way of doing it. I ran a preliminary pin-roller thrust calculation with a similar setup in a different project, got a horizontal deflection value that felt too high (which goes against what you said about expecting a small value), and just added a horizontal member because it was a legit (but frowned upon) option in that project. The horizontal member is not really an option in this one. I think I'm going to exercise my right to design at least one thing in my life based on my gut, and do some quick calculations and call it a day, and rely on other parts of the structure to take care of it. Because it's not just a huge window in isolation. I also feel like wood calculations aren't that precise compared to steel or concrete, so I'm not going to overthink it.

@everyone I got some fantastic ideas and comments here, thank you all. Not saying that the thread is over, in case there are other ideas. Just that I'm very impressed and appreciative of how people think over here.

 

[Eng16080]

Here's another idea with two cantilevered beams supporting a structural ridge beam at the peak. For good measure, I would add a tension strap between the beam below the window into blocking in the tall wall segment. This will help provide continuity between the tall wall segment and the portion below the window. Although the beams are designed as cantilevers, the shear walls will provide some additional support/restraint in terms of resisting outward thrust. Be sure to check the deflection at the cantilevered end to ensure it works with the window unit. Possibly sandwich a steel plate between LVL plies if additional stiffness is needed.

Below is a lousy sketch to illustrate:

 

[SE2607]

XR250 - Regarding "The ridge is supported vertically by a bent flitch or I-beam at the gable end - thus eliminating thrust. Similar to this..", I'm not having any difficulty understanding what you propose. It is a very common approach. In fact, I've done this several times. However, once I modeled this framing in 3D, I was really surprised about what is really going on. Consider this: visualize the ridge beam deflecting (which it will). What happens to the roof rafters? Do they compress axially to compensate? They do not because the axial stiffness of the rafters is greater than the flexural stiffness of the ridge beam, even if you use a GINORMOUS ridge beam. The ridge is either going to be supported by the rafters, transferring the thrust to the adjacent structures, or the walls are going to bow outward, resisted only by the flexural/axial capacity of the top plate (yikes!).

Milkshakelake - When you are visualizing the plywood sheathing taking flexural loads, please consider that the boundaries of the individual plates (ridges, hips, etc.) are continuous pins (think piano hinge) and not able to transfer moment.

My recommendation still stands, and I think, with some assumptions, the analysis and detailing is not that onerous.
1. Use a 2x ridge board. Anything more is a waste of lumber.
2. Analyze the rafters as a pinned-pinned truss.
3. Apply the horizontal and vertical reactions to the adjacent structures. Note that the thrust load in reality is not linear, but parabolic, with zero thrust at the ends and maximum at the middle. To convert to an equivalent uniform load, use 2/3 of the maximum load, but design the adjacent rafters for the maximum load.
4. Connect the rafters to the adjacent structure with a sleeper and blocking between the supporting rafters.
5. Consider this load in addition to the RLL+DL when designing the rafters on the adjacent structures.
6. Take the lateral load out in shear resisting elements of the adjacent structures, shear walls for the element on the right of the entry, cantilevered columns on the structure to the left of the entry. Depending on the load, you might be able to get away with wood posts and Simpson MPBs. Be careful of the load factors. The capacities for the MPs are based on 1.6 load factors. Also be mindful of the deflection. This is a long term load and wood is subject to creep. This element is visually exposed, so any significant deflection will not make you look like a hero in a couple of years. HSS columns, wrapped with wood are relatively cheap. You probably can get away with a 2 bolt base plate as it will weigh less than 300 pounds. The forces shouldn't be huge. Use flagpole footings restrained at the top by a tie beam (footing).
7. You're done! You had some fun! You don't have to think about this ever again!

 

[XR250]

XR250 - Regarding "The ridge is supported vertically by a bent flitch or I-beam at the gable end - thus eliminating thrust. Similar to this..", I'm not having any difficulty understanding what you propose. It is a very common approach. In fact, I've done this several times. However, once I modeled this framing in 3D, I was really surprised about what is really going on. Consider this: visualize the ridge beam deflecting (which it will). What happens to the roof rafters? Do they compress axially to compensate? They do not because the axial stiffness of the rafters is greater than the flexural stiffness of the ridge beam, even if you use a GINORMOUS ridge beam. The ridge is either going to be supported by the rafters, transferring the thrust to the adjacent structures, or the walls are going to bow outward, resisted only by the flexural/axial capacity of the top plate (yikes!).

Umm, are you in academia?
Stop modeling this in 3D. The ridge deflects, the walls move outwards slightly - big deal. Every roof does this to some degree - even trussed roofs or standard rafter/ceiling joists framing deflect outwards due to elongation of the bottom chord/joist. Scissors trussed roofs are even worse.
Using the adjacent structures to resist the thrust is an option - but one that will never be implemented properly and will require a huge amount of detailing. I'm in this business to make money and keep clients. How about you?
I have seen the adjacent structures work to effectively brace a roof like this - but it was not intentional. It worked fine though.

 

[SE2607]

XR250 - Nope. Not in academia, but I have studied the secondary effects of modeling in 3D. Deformations due to flexure are MUCH greater than deformations due to axial loads.

You are welcome to ignore them. It doesn't matter to me.

 

[SE2607]

I might add that the solution I presented is the most cost effective and defensible. Ridge board, Norridge beam. No inverted steel chevrons. Just two flagpole blcolumns at the entry, which may be wood, depending on the loads.
The calculations aren't too onerous, particularly if one uses 2/3 of the maximum thrust as a uniform load and applies the total thrust to both adjacent diaphragms instead of distributing proportionally based on relative stiffness.

My caution in general is to at least consider deformations with any structural analysis and apply proper boundary conditions, or at least envelope all probable boundary conditions.

Remember, just because you call a tail a leg, that doesn't make it a leg.

 

[phamENG]

Sorry I missed this one. Not sorry I was on vacation.

While I agree with the validity of SE2607's approach and like the rigor of it...I have to side with XR250 in practice.

After all...the rafters may not shorten enough to matter, but at free edges the walls and the beams being bent about their weak axes will be moving. And while we generally simplify wood diaphragms into a 'flat plate' for wind loads despite hips and valleys...there's a difference in doing that for short term wind loading or seismic loading and doing it for long term dead load or roof live loading. The flat plate assumption assumes that it's adequately supported. If we are asking the diaphragm itself to resist gravity forces like this, we're getting into folded plate theory and design. Doable, yes, but also very cumbersome and the loads you end up with can be huge. I haven't used it in practice yet, and I don't want to, but it's a trick that I know is out there. I don't want to do it because the detailing is abnormal. Nothing invites problems on a house build more than something that is 'abnormal.'

In my area the 'flagpole columns' would have to be steel. No chance of getting wood to work without embedding it, which is just inviting rot and termites in 5-10 years. Sure, with the right treatment, blah blah blah...but it'll be a 6x6 off the rack at home depot as likely as not. The cost of that plus a moment resisting footing (which, again in my area, would have to be a shallow spread footing as you'd hit water before you had sufficient embedment with an augered sonotube type foundation) would be quite high. Much higher than a beefed up cranked flitch beam.

At the end of the day, I find these jobs to be a balancing act, with a couple absolutes:

1) Ensure that there is a stable and reliable load path for design level loads to prevent collapse.
2) Ensure that the 'alternate' load paths that are almost invariably present and may be stiffer than my intended one will not cause damage at service level loads.

While I want to understand the behavior as much as possible, I'm in this to make money. Is my design safe and serviceable? Can the contractor build it with limited fuss? Great. Time to move on. If I'm re-inventing the wheel to make sure I know that my dog only has 4 legs...then I've taken it a bit too far.

 

[SE2607]

If the loads are low enough, you can use wood posts with Simpson's PMBs.

O.K. I got it. Ignore boundary conditions and deformations because it would prevent me from making money since the analysis would be so onerous to justify.

Got it.

 

[XR250]

If the loads are low enough, you can use wood posts with Simpson's PMBs

What is a "PMB"
Also, I think you will find that the outward deflection in a wood post will be significant in most situations of this scale.
I mean, you could use steel posts and tie them together below the window to get a propped cantilever and remove the moment on the footing. (just noticed Jayrod already suggested something similar)
By the time you do all that, might as well do the flitch as it will be cheaper and easier.

 

[driftLimiter]

A decent ridge beam is gonna have me sleeping a lot better than a 2x ridge board and some cantilever wood columns I can tell you that.