Truss Modeling

Today, I model it as I intend for the fabricator to build it.

Years ago when we calc'd everything by hand, we assumed all the chord intersections were idealized pins. The moments were refered to as "secondary moments" and were generally ignored in the design if the truss was loaded at panel points only. We would include bending moments however, if the chords were loaded between panel points.

I was always told to model the chords as continuous and the web members as truss members (in STAAD), because that is the way it is going to be constructed. I tried to argue that by modeling it that way was conservative because of the moment that was induced in the chords...but I always lost. So, I don't know if that is the right way but it is the way a lot of people model it.

remember at the truss nodes that several members meet (some maybe continuous) so that there is a bunch of load redistribution happening at the node (presumably you've got a gusset connecting the members together). your model welds all the elements together at a node, which isn't the real geometry, so you need to check that the loads can shuffle round the way the FEM wants them to go.

if you wanted to model an ideal truss, you could model each span as a single rod element (rather than a beam) ... this'll produce slightly higher loads in the members, but it'll miss the higher loadeds on the joints (due to the fixity that occurs in your design (particularly with the continuous elements).

I vote for modeling exactly as it is going to be constructed. Show pins only where the chords are going to be spliced. I am all for modeling as close and as tight as you can, then add conservatism so you can measure how much conservatism you actully have.

Why are you getting a lot of moment in the chords? If the configuration looks like a truss (ie all triangles) with no eccentric connections, the only moment should only come from the loads you have applied to the members them selves. You might want to check your model. I am not sure how STAAD works, but RISA will give you a warning and lock nodes if you have not modeled it properly.

I believe either method is adequate and permissable as both methods produce statically admissable structures as long as the members are designed for the forces that are assumed. It is really up to the designer.

Method 1: Chord continuous with pinned web members - will give you moments in the chords and slightly less force in the web members. Chords must be designed to handle the moments. A reduced effective unbraced length factor could also be used for axial capacity of the chords (say around K=0.8) considering the joint fixity.

Method 2: Everything pinned. Still statically determinant. No chord moments - higher web forces. In order to get to this condition the members of the actual chord at ultimate load will have to go plastic at the web joints - thats fine as long as they are of a ductile material (such as steel). Chord axial strength would be based on K=1.0 of length between panel points.

I design heavy timber trusses for a fabricator. If a member is continuous across the joint, I all ways model as fixed at the joint.

just a little pic at willisV's post ...
a "true" truss (with endload members) can be redundant