Pinning Connections, End Releases

[tovins76]

I am currently modeling a 9 truss hangar and am having some problems with somewhat large deflections and very large member stresses. I currently have the end boundaries pinned and have also tried having them completely fixed as well as free. I wasn't sure if anyone has any idea how the truss boundary conditions are usually modeled in Risa? We are working with older steel so we like to see stresses under 33 ksi and have gotten some as high as 45 ksi which is not good. Any help would be appreciated. We are analyzing dead, snow, and wind load if that is any help.

Attached is a picture of the design thus far.

 

[JoshPlumSE]

It's hard to tell without having the actual model. But, based on that picture I have a few guesses:
1)You've got some plate elements that seem to represent the end wall or the doors of the hangar. These are not meshed properly. Plates need to be sub-meshed into smaller pieces in order to provide accurate resutls. Specifically plates only lconnect to their 4 corner nodes, they don't have connectivity to joints or members that run along the interior of th plates. Therefore, you want to sub-mesh the plates such that you have corner joints everywhere that you want to connect.

2) You've got some triangular plates on the side walls. My guess is that these are intended to act as some form of stiffener bracing the columns for translation or such. Same sort of issue applies there. The plates need to be sub-meshed. I'm also not sure how the boundary conditions in this region are set up. That may be problematic as well.

 

[tovins76]

Ok thanks, the sub-meshing seemed to help a bit. The problem areas still seem to be on the truss members themselves. In Risa, are nodes on a truss assumed to be pinned, free, fixed, etc. I think this may be a problem along with the I and J releases. Any help with those two areas would be greatly appreciated.

 

[tovins76]

Also, when the members are pinned we are getting sensible stresses but zero deflections which can't be right. Then when free we are getting both stresses and deflections that seem much too high.

 

[JoshPlumSE]

In RISA, we do not have a specialized "truss member". Therefore, these members are capable of taking bending and axisl force. You can use the member end releases to ensure that no moment develops at the end connections of those members.

When you say that the deflections are zero when the end is pinned, it sounds like you may have accidentally used a pinned boundary condition rather than a pinned end release.

Looking at your model, I am wondering what sort of retraint there is to the transverse buckling of the truss. There does not appear to be any X bracing in the roof system to prevent this type of buckling. Therefore, I could see a P-Delta analysis really killing these trusses. Sure, you've got some joists transverse to the trusses. But, if those joists have pinned ends then they're not going to offer much lateral resistance.