Square Hip Roof with Glulam Hip Beams and Rafters - Will a square tension ring work?

[spencerltu]

Hi all, first post here but certainly not the first thread I've read.

I'm designing an open pavilion approximately 65'x65' with center peak. Roof slope is 4:12. 12 perimeter columns HSS 12x12x1/2 with perimeter beams HSS12x6x1/2 at elevation 10'-0" to pick up the rafters. The glulam design (by others) shows 4 hip beams (36" deep) to the corner columns. The thrust of these hip beams is 89 kips (890 kip-ft moment). My concern was finding a column to transfer that force to the foundation. With that kind of moment, the only column that would work was a very heavy W14.

After speaking with the glulam designer, he said the thrust would be transferred into the perimeter beams as 63 kips tension and they would act as a tension ring. cos^-1(40/89)...I guess. If that's true, the columns will only see the axial force. I'm having a hard time convincing myself that the column won't see ANY of the thrust force.

I can understand the theory, but have any of you ever used a tension ring in a square application rather than circular? This designer has many years experience on me and says he has done it many times. I'm just trying to get comfortable with his design.

Any thoughts/comments are welcomed and appreciated.

Thanks all!

 

[KootK]

I think that the concept is valid for the portion of the applied load that is symmetric (i.e. the dead load). Any unsymmetric loads will result in distortion of the square and thus either a) in plane moments at the knuckles of the square and/or lateral loads on your columns. This is also true of a circular ring of course. A circular ring is just better suited to handle the distortion because of its geometry. Is there a competent diaphragm of intermediate framing that might be able to help out with the asymmetric loads somehow?

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[BAretired]

A square arrangement of tension members is not stable by itself as it is not composed of triangles. You need to add a diagonal member in order to render it stable.

BA

 

[KootK]

I sketched up a few free body diagrams and I think that I'll have to change my answer. I think that the arrangement will work for unsymmetrical loads as well. What it comes down to for me is this:

If the roof were flat, you'd have a fully cross braced, eight member shear panel incapably of deforming into a parallelogram. This is the triangulation that BA is looking for. Even with the pitch added to the ridge beams, I think that this is still true. The connections at each end of each ridge beam will need to be able to transmit both shear and axial forces into the joints. That shouldn't be too tough to manage however.

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[msquared48]

I have seen the concept work and designed two or three myself.

The rigidity if the roof diaphragm plays a huge part in the stability of the structure in conjunction with external shear walls.

Mike McCann, PE, SE (WA)

 

[dhengr]

Spencerltu:
I agree with KootK, BA and your designer, this is a perfectly reasonable framing arrangement. But, the devil’s in the details. For the most part, the four corner columns will only see vertical gravity loads from the roof above. The vert. column loads may be different on one side than another, same for the thrust loads, due to unsymmetrical loading, but this is all resolved at the eave beam level and through the roof plane diaphragms. The eave beam, tension beam (don’t call it a tension ring, if you wish) will take the thrust from the hip beams, as tension at the eave elevation. The columns don’t see moments from this loading. But, this structure needs column moments, some form of bracing, etc. for lateral loads like wind or EQ loadings. And, the structure can just lean over, in any direction without this bracing system or moment cap’y. These structures are also very susceptible to a mode of torsional collapse, where they tend to rotate about their vertical central axis.

The eave beams, tension beams, must also take vert. loads and lateral (horiz.) loads from the roof rafters. Just as on a simple gable framed roof, there is potential for lateral thrust out at the exterior bearing walls (the eave beams in your case) and this must be accounted for: either through lateral strength of the eave beam; the roof plane diaphragms deflection (lack thereof); or a few tension ties across the bldg. at the eave elevation. Despite protestations about the tension ties, you still get a very open roof system, and they are great places to hide wiring, hang lights, fans, speakers, etc. and can even be used to support a fire pit hood and chimney. You also need a compression ring of some sort up at the top to facilitate the connection of the four hip beams. Since this is a square hip roof, there will be no ridge beam.

 

[XR250]

@ Dhengr;

There should not be any lateral thrust from the rafters as the hips are acting as beams to vertically support the high end of the rafter.

 

[Triangled]

BA's note makes me think of unbalanced live load conditions, and the solar panel salesman wanting to stack the south facing panel, "absorbing" live load, a few years hence.

 

[dhengr]

XR250:
That’s sorta true, except when its not. That is what we’re taught and generally assume in basic house design. But, unsymmetrical loading will very likely lead to some funny thrusts through the roof system, certainly wind and EQ loading will. And, when the hip beams shrink a bit in length, and the two compression end joints settle into place, and the roof peak settles a little bit vertically. Then also, the rest of the roof framing may settle a little as well, in its bearing hangers as they tighten up. The movement of the rafters also had to do with whether the two bearing seat cuts are horiz. or if the top cut is a vert. sloped cut on the rafter. And, the sum of these settlements leads to some lateral movement and thrust out at the eave beam. An eighth of an inch vert. settlement up near the peak translated to about a quarter inch or more of lateral movement at the eave beam. You figure out how much and what affect, just don’t forget that possibility. The roof plane diaphragm may help mitigate this issue, and ceiling joists (ties) properly attached would help, but not likely be welcomed. You see this lateral thrust and movement often in houses and other long gable bldgs., not much, but it is there, and it is not assumed to be there, until it is.

 

[XR250]

@ Dhengr;

My guess is the thrust is self limiting and dies not really need to be designed for. The roof diaphragm should pick up the slack.

 

[RFreund]

@BA - I think I see what you are saying but can you expand on that? In this case, given the diaphragm I'm not sure that is applicable. Or maybe it is.

EIT

http://www.HowToEngineer.com

 

[BAretired]

@RFreund - What I am saying is that a system of four members in the form of a square, hinged at each corner is an unstable structure because it can deform into a parallelogram. This is true even if all four tension forces are equal.

The structure can be made stable with the addition of one diagonal member joining two opposite corners; or by designing the roof diaphragm for the worst case of unbalanced load; or by designing the columns or shear walls to carry lateral forces which may arise from gravity and wind or seismic load combined.

It is a mistake to assume that the presence of a diaphragm and four columns will necessarily be sufficient to account for all possible combinations of load; it might and it might not; it has to be investigated. It is a disservice to advise a young engineer that everything is A-okay without doing some further analysis.

BA

 

[XR250]

@Ba;

Seems it can only go into a parallelogram if the elevation of the top of two hips changes in relation to the other two hips - which cannot happen - unless somehow two hips get longer and two get shorter.

 

[msquared48]

BA: Shear walls or frames at the sides will prevent the rotation you describe. Done all the time.

Mike McCann, PE, SE (WA)

 

[Canuck67]

A "square" ring beam can be made to work for smaller structures, not sure of whether it can be made to work for the loads you have on your size of structure. The perimeter beams need to be checked for both vertical (gravity) and horizontal (thrust) loading. Then the hardest part is making the perimeter beams rigid at the corners. Full moment connection to resist the horizontal thrust forces and prevent the square becoming a parallelogram. Or perhaps horizontal corner braces at each column location between the perimeter beams will help. Without a ring beam or other potential option, the horizontal thrust resulting from the gravity loading will be problematic. Horizontal thrust will be present unless the peak is somehow fully supported to prevent any vertical movement.

 

[KootK]

Seems it can only go into a parallelogram if the elevation of the top of two hips changes in relation to the other two hips - which cannot happen - unless somehow two hips get longer and two get shorter.

I believe that this is the crux of it. And this is why I changed my answer to "all good" at the top. The ring is cross braced, even though that cross bracing is pitched. I also did the statics for an arbitrary inclined point load applied along one of the ridge beams and everything was internally resolvable.

I'm convinced that the arrangement is stable for all loadings and does not require the assistance of either diaphragm action or moment connections at the ring beams. Of course, lateral loads still require a path to the ground.

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough to either change it or adopt it.

 

[BAretired]

@XR250;

Yes, the hip beams, together with the four tension members at eave level form a stable three dimensional structure if adequately connected. Under uniform snow load over the entire roof, each column carries one quarter of the total load and the vertical shear in each hip beam is zero where they meet at the peak.

Under unbalanced snow, the vertical shear in each hip beam at the peak can be non-zero, so the connection must be adequate to transfer that shear. It may seem a trivial point, but assuming the system will work without checking the central connection for shear could lead to collapse.

BA

 

[RFreund]

@Canuck67 at first I thought there would be lateral thrust on the perimeter beams as well, but there would not be if it is all taken by the hip beam. If the Hip beam were "non-structural" or if you had a "non-structural" ridge, then yes there would be lateral thrust which would need to be taken by weak-axis bending of the perimeter beam. Also I don't think the connections need to be fully rigid.

EIT

http://www.HowToEngineer.com

 

[XR250]

Under unbalanced snow, the vertical shear in each hip beam at the peak can be non-zero, so the connection must be adequate to transfer that shear. It may seem a trivial point, but assuming the system will work without checking the central connection for shear could lead to collapse.

Agreed. Something that I prolly would not have thought of. I have never done one this large. For a smaller one (25' or so), whatever reasonable connection that gets used and the diaphragm likley just make it work.

 

[BAretired]

Another point to consider is erection of the hip beams. Until all four hip beams are in place and connected, the structure is unstable. The hip beams are installed by a different crew than the edge beams and columns because they are glulam, not structural steel. One way to do it would be to install a temporary central column, then remove it after all hips are installed. Another would be to brace the corner columns so that they can take any lateral forces resulting from temporary unbalanced dead load.

Before the steel crew leaves the site, they have to ensure that their part of the work is secure. To do that, the columns have to be braced or fixed at the base.

BA