If it is a two dimensional frame subject to a large degree of torsion, which I would consider rare, I would not release the torsional moment as it will increase the shear to the anchor bolts.
If it is a three dimensional frame with significant torsion, which would be more common, I would have to think about it. I might model it both ways here and look at the difference.
Sometimes - like with RISA you need to leave it "on" so it doesn't fall over. If truly two dimensional - the loads will be just a couple of pounds - but RISA won't give a load of errors about instability...
When I model fixed beams, I assume they will be CJP welds, which, in my mind, is fixed torsionally, therefore, I model it that way. That being said, I don't think that you'll see any difference in your frame results whether you fix or release the beams torsionally.
To me, virtually all moment connections are torsionally restrained. They may be torsinal "pins" without warping restraint. But, they are torsionally restrained.
I'm not sure how we could achieve moment fixity at the ends of the member without having some level of torsional resistance as well. Maybe one of the partially fixed moment connections which are supposed to be pinned for gravity loads, but fixed for lateral loads? I've never really use those types of connections, but that would be the only way I could see releasing torsion for a moment connection.
Do most of folks out there use "beam offsets" when modeling moment frames? Say you have a moment frame with two columns and one beam and the beam is framing between column flanges, do you offset the beam and in order to account for dead load moment in the columns due to the connection eccentricity?
It amazes me how many people ignore this effect in frames that dont see balanced loading.
@ToadJones - I ignore it - in my opinion this is not any different than ignoring the eccentricity to the column in standard shear connections which is done all the time. The increase moment is generally offset by the added rotational restraint that the connection gives to the column that is generally ignored in column design.
I guess I dont ignore it on shear connections either unless they are balanced on the other side of the column or if they are connected to the column web.
I have had a few projects where the connection eccentricity and the shear loads from massive girders created a pretty significant effect on the columns.
I always ignore that "rigid offset" for moment frames. It doesn't significantly end up increasing the moment in the columns.
IMO, the biggest argument for using those offsets relates to reducing the drift of the frame. But, in seismic you should not normally use it because (for steel frames) you are probably also ignoring panel zone shear deformation. According to NEHRP commentary, if you ignore the rigid region, then you can also ignore panel zone deformation.
For shear connections, it's a more interesting question. I usually feel perfectly comfortable ignoring the effect for interior frames. For exterior columns, it's not such an easy call... though I still usually end up ignoring it. I just might take a closer look at my column size, my connection eccentricity, my total shear and my column footing support.
If I do want to consider it then I just put in a rigid end offset like you suggested. Odd that rigid end offsets were originally intended for moment frames, but current structural practice makes using them for shear connections more common.
"
IMO, the biggest argument for using those offsets relates to reducing the drift of the frame. But, in seismic you should not normally use it because (for steel frames) you are probably also ignoring panel zone shear deformation. According to NEHRP commentary, if you ignore the rigid region, then you can also ignore panel zone deformation. "
For moment frames rigid end offsets end up reducing the frame stiffness a bit while the panel zone shear deformation ends up increasing the drift a bit. The NEHRP commentary essentially says that the two effects end up canceling each other out for most structures.
That Panel Zone Deformation section of the code has always bugged me. If the commentary is going to say "this effect can be ignored" then how / why does a provision like this ever makes it's way into the code?!
Again - I tend to ignore the eccentricity of connections to columns due to the restraining effect they have on the column. See first sentence of Charlie's answer here: