I am modelling a very long lintel beam carrying a low roof, masonry, and a higher roof. Roof structure is composed of open-web joists. The low roof joists will be seated on the bottom flange. To model torsion induced by the slight eccentricity of the joist seats I am using a very short plate of the flange thickness, offset by rigid links to the elevation of the bottom beam flange. All of this works great and I get good, reasonable results with expected torsional moments, vertical and horizontal deflections. My problem arises when I use the very same W-shape in either A992-50 or A572-50 steel. The A992 shape results in a 2% overstress (102% usage), the A572-50 shows a max usage of 54%, with torsion governing the result. Even A-36 has a max usage of 91%.
I am using the materials supplied with the program, not custom materials. The only rational explanation I can find for the result is the Poisson ratio for A992 is shown at 0.3, and 0.26 for the other steels. Is this accurate?
Which program, Ram Elements or STAAD.Pro? Are you checking torsion per design guide 9? Can you send the file?
Elements. I've replied via email and uploaded the files.
There are some custom sections in your model that I don't have. There is also an instability when I try to run it in version 16.03.00.1010. What version are you using?
I replaced the sections and restrained one end of the beam in torsion in order to get results. After doing that I get the same design/capacity ratio using A992 or A572. The variation in Poisson's ratio could account for some slightly different deflections, but I would not expect it to affect the torsional demand or design for your model.
It does appear that 0.26 would be the more accurate value of Poisson's ratio.
I am running 22.214.171.124. The custom sections are simply ¾” thick plate 6 wide. I am not restraining the ends to model a pinned end for torsion; the beam will be in masonry however not restrained so I think that is safest, and closest to the available nomographs for torsion, however there was no instability in the runs I had.
The Poisson’s was the only variable I could observe, and therefore assumed with the resulting variation in deflections eccentricity was changing and resulting in more torsional stress.
I have uploaded the custom section and a screenshot of it. Clearly this lintel is carrying a lot of load and I want to be very sure, if you would please take another look with the above information.
Make a new A992 material in your version (see: https://communities.bentley.com/products/ram-staad/w/structural_analysis_and_design__wiki/3569/creating-custom-sections-materials-etc-in-ram-elements-tn ) and use 0.26 for Poisson's ratio to rule that out as a factor.
Otherwise update to v16.03. I see at least one issue that was fixed for torsion design in the release notes.