Truss Design

[jcox]

The majority of the roof systems that I have designed are flat roof systems, and for th most part are pretty easy to design once the loads are established. One thing that has always bothered me is sloped truss design. I know that joist designs are assumed to be pinned/roller support condtions. I have modeled some trusses where the support condition is pinned/pinned, and pinned/roller. The pinned/pinned condition usually induces axial forces high enough in the chord members to produce member failure. I assume the pinned/roller model is the most common model used. If so, what kind of connection do you use to accomodate the horizontal displacement of the roller end. Then if you assume a roller end, how is lateral load accomodated? Do you suddenly assume a different support condition? I realize that the supporting element probably "flexes" enough to accomodate the displacement. Its just the idea that we design it as not connected but connected that really bothers me. To me this is similar to designing a curved beam as a beam, connecting both ends to a wall, and expecting no arching action.

 

[Kramer]

Trusses are typically supported by bearing walls at each end. A typical wall can accomodate quite a bit of sideways movement without distress (if you have ever framed a wall before you have a good feel for this). For flat bottom trusses, the amount of movement that you need to accomodate for the pinned/roller connection is quite small. Scissor trusses however can spread quite a bit laterally, depending on the roof and ceiling pitch. You have to apply engineering judgement in deciding how much spread is acceptable. One consideration is what effect this spreading will have on the walls parallel to the trusses.

 

[Trussdoc]

In the typical scissor-type wood truss (ie metal plate connected 2x wood, two feet on center), the spread due to lateral thrust is counteracted in two ways:

(1) Proper installation of metal clips with slotted fastener holes that permit some movement of the truss without transmitting it to the supporting wall. Once dead load is in place, the walls can be plumbed again if needed, and the only lateral movement will be due to live or snow load.

(2) Although calculations on the truss itself may show an alarming amount of horizontal deflection, the reality is that the roof diaphragm, generally plywood or structural panel, acts as a long deep beam, anchoring the trusses to the end gable walls or to the hips. For a typical US residential or small commercial building, the summed lateral forces from the trusses create a movement of less than an inch at the mid-point of this "beam".

ALSO: Using "pinned-pinned" design for a scissor truss results in much lower member stresses and deflections than "pinned-roller" analysis. Running a quick example gave me a difference of well over 100%!!! Never approve a design using more than one pinned bearing until the capacity of the bearings can be thoroughly justified.

 

[Kramer]

Trussdoc,
I agree with your comments. I do have one question though. I am typically the EOR for the overall structure with the trusses being "design by others". I do not specify the slotted connectors at the supports on scissor trusses. The reasons are that for proper shear transfer from the roof diaphragm to the top plates, I specify a shaped 2x solid block with a framing clip to the top plate. This 2x solid block is specified to be toe nailed to the trusses at each end of the block. These blocks are put in prior to the installation of the roof sheathing. Obviously almost none of the dead load is in place at this point. The shear blocking will not allow the truss to slide on the support rendering the connector ineffective. The other consideration is that there needs to be resistance to wind loads against the face of the wall. The slotted plate connectors do not provide any such resistance.