Are joist chords analyzed as continuous or pinned at panel points?

[SeizeTheMoment]

The joist calcs I have usually reviewed from manufs usually only include axial loads in their member forces. I haven't seen a calc that shows how moment along the chords are considered.

This leads me to think of a couple ways of analyzing a joist:
1) Case 1: Everything is pinned, and every load is a point load only at the panel points. This is my best interpretation of how there would only be axial forces in the joist members, but I don't believe this is a real life scenario.

2) Case 2: Everything is still pinned, and loads are applied as a distributed load. This would include moment at least for the top chord in a gravity loading. Unless there are purlins that are connected to the panel points, I feel like this is the case for most scenarios.

3) Case 3: Chords are continuous, there are moments on the chord at panel points. I think (?) this is how it would realstically behave, but I know this is stepping away from how truss stresses resolve and more into beam theory...


My understanding is that joist calcs usually only consider pure axial loads in joist members, leading me to believe they consider Case 1. I think there's a gap in my understanding with how the chords are actually considered at panel points (whether continuous, or pinned), and how loads are applied.

Let me know your thoughts, thanks!

 

[ANE91]

Secondary Stresses in Trusses by Shankar Nair is a good starting point for this topic.

The short answer is that you have to decide what the most appropriate analysis is, based on member stiffness and node fixity. Mandatory point loads can dominate the structural response when sufficiently large and located between panel points (e.g., RTU). I usually model chords as continuous and nodes as moment-released, but there are certainly situations where you must consider panel-point rigidity. Vierendeel truss is an extreme example.

In reality, everything bends a little.

 

[ANE91]

Also, what software permits releasing every end at a node without throwing a mechanism error? I’ve always gotten yelled at for that…

 

[SeizeTheMoment]

Also, what software permits releasing every end at a node without throwing a mechanism error? I’ve always gotten yelled at for that…

This is RISA, and yes there is an error. I was just investigating how it would distribute the forces.
In the analysis that you have used, are there negative moments at the panel points if you consider a chord to be continuous?
I'm curious how moments are included in the D/C ratios, usually I only see the axial force and a D/C ratio, so I'm not sure how any moment has been considered.

Here is a sample:

 

[RontheRedneck]

"Case 3: Chords are continuous, there are moments on the chord at panel points."

That's how trusses are currently analyzed.

 

[ANE91]

are there negative moments at the panel points if you consider a chord to be continuous?
I'm curious how moments are included in the D/C ratios

Yep, and this is where things get tricky. The chords go into bending. You’ll need to pay close attention to unbraced lengths, so that your chords (and webs) don’t buckle. RISA has some nice shortcuts for inputting the various stability factors. It gets a little confusing deciphering which buckling mode (e.g., flexural vs lateral-torsional, etc.) is under consideration by the software. Codes handle the unbraced length issue differently. For example, TPI allows 80% of the node-to-node length, in most cases. SJI probably also has documentation on this.

If you chase this particular rabbit deep enough, then you’ll find that the bottom chord needs to be braced even when in tension. (Spoiler: the bracing is for the compression diagonals.) This is not intuitive, and the truss/joist is usually OK as is, but it’s worth checking. Refer to The Importance of Tension Chord Bracing by James M. Fisher.

 

[XR250]

Also, what software permits releasing every end at a node without throwing a mechanism error? I’ve always gotten yelled at for that…

Why would it throw a code? Is it somehow unstable.
My software lets me do it if I am understanding your statement correctly.

 

[Celt83]

Why would it throw a code? Is it somehow unstable.
My software lets me do it if I am understanding your statement correctly.

If all member ends are released at a node then the node itself is free to spin, puts a 0 on the diagonal of the stiffness matrix. In the software I’ve used they either throw an error or warning about this and Risa will usually lock the node and continue the run.

 

[XR250]

Hmm, I guess I don't understand what "end released" means. To make a pinned structure in my software, I can either make a node a "pin" or I can make the end of the each member not transfer any moment - same results. Granted, I am using 2D software from 1985.

 

[Greenalleycat]

Hmm, I guess I don't understand what "end released" means. To make a pinned structure in my software, I can either make a node a "pin" or I can make the end of the each member not transfer any moment - same results. Granted, I am using 2D software from 1985.

More or less the same I'm sure
In my software, SpaceGass, I can get zero moment nodes easily enough but it's all done via 'making the end of each member not transfer any moment' to use your lingo
However, if I make all members at one node 'not transfer' moment then it will throw a hissy fit and error
So, say I have 3 beams meeting at a node, I'll release the moment on 2 and leave one moment transferring
The end result is correct (no moment transferred) but it's required in the background for the analysis to run

 

[Just Some Nerd]

In my software, SpaceGass,

Definitely gets a little annoying for me too when trying to analyse a 2D situation in 3D software and having to add restraints everywhere