1. by rule of thumb I like to frame beams at least 1/2 depth of supporting girders (I don't see literature on this)
IMO, this isn't necessary. Not bad as a rule of thumb for preliminary sizing and such. The key for me is that we want to brace the spandrel beam for LTB (lateral torsional buckling) which really requires that the compression flange (top flange) of the spandrel be braced. Now, if the spandrel is also taking AXIAL load like a strut, then 1/2 depth beam may be required (to prevent weak axis buckling of the spandrel).

2. I commonly see min. 3 bolts for spandrel connections (I assume for some kind of moment capacity).
Are you talking about how the spandrel connects to the columns? Or about how the floor beams connect to the spandrel? Two bolt beam connections are fine for the small redundant / repetitive joists. I generally see the minimum # of bolts spec'ed out based on beam depth. Just easier for contractors to keep track of that way. We might end up being a little conservative in some cases, but the simplicity is worth it.

3. I have always seen full depth plate connections for floor beams to spandrels. I assume to prevent torsion, but I cannot find substantial literature on this either.

If the spandrels have moment connections at their ends, then the bottom flange will go into compression. Since you don't have beams and decking to explicitly brace the bottom flange, we might use full depth stiffeners on some of these beam to spandrel connections.... to limit the LTB buckling when the bottom flange is in compression. But, we don't have to do that full length. Just a little bit beyond the point of inflection.

Also, in other cases (where you can't easily use a beam connection) you can add in explicit braces from the deck down to the bottom flange.

Not quite sure...

1. by rule of thumb I like to frame beams at least 1/2 depth of supporting girders (I don't see literature on this)
often they are and sometimes a little deeper, but not really necessary.

2. I commonly see min. 3 bolts for spandrel connections (I assume for some kind of moment capacity).
On a major project in Saskatchewan that I did about 30 or so years back... W14x22s were attached to W24s with two bolts and single angle connections... if memory serves... This was the first project I did with single angle connections... and there were 1000s of connections... made a bit of an oversight, but that's another story... I should have issued the change as a Change Order...

3. I have always seen full depth plate connections for floor beams to spandrels. I assume to prevent torsion, but I cannot find substantial literature on this either.
I almost never use full depth plates... usually stop 1" or so from the top flange... it's cheaper and easier to fabricate. In the project above connection was by clip angles.

Rather than think climate change and the corona virus as science, think of it as the wrath of God. Feel any better?

-Dik

Quote (OP)

1. by rule of thumb I like to frame beams at least 1/2 depth of supporting girders (I don't see literature on this)

In the absence of a more detailed investigation, it is commonly assumed that a roll beam should be about 60% of the "T" dimension of the beam being braced. I'm pretty sure that I've also seen this in print in some AISC document at some point in time.

Quote (OP)

2. I commonly see min. 3 bolts for spandrel connections (I assume for some kind of moment capacity).

That probably does speak to roll beam bracing and/ or torsional capacity but I'd also be fine with two bolts as that also gets you a nominal couple. Most of the time, I just see as many bolts as will fit at 3" oc.

Quote (OP)

3. I have always seen full depth plate connections for floor beams to spandrels. I assume to prevent torsion, but I cannot find substantial literature on this either.

Again, as you've anticipated, this suggests an attempt to create a robust torsional connection to the infill beams, either for the purpose of torsional bracing and/or for resisting direct torsion.

Quote (OP)

I am afraid that the by the time my floor beams meet all the above criteria (cope 4", provide three bolts, 1/2 depth of spandrel + 4" drop min beam depth) the client will wonder why infill floor beams with modest spans might be so deep.

In the situation that you're describing I do not feel that it would be reasonable to incur a large weight penalty just to satisfy the [0.6 x T] psuedo-requirement. Rather, I feel that many engineers would attempt to accomplish the same thing with kickers installed to connect the bottom flange of the spandrel beams to the top flanges of the infill beams. It creates the same roll beam connection but, in some cases, at a lower installed cost. Of course, the kicker has to be palatable from a spatial perspective as well and not stick out through ceilings etc.

Where full height stiffeners are used in conjuction with a deliberate moment connection, that may well be an attempt to provide torsional bracing / capacity in a situation where diagonal kickers would be unacceptable. Appendix 6 of AISC SCM provides the requirements for strength and stiffness in such situations for the bracing function.

Quote (JP)

Now, if the spandrel is also taking AXIAL load like a strut, then 1/2 depth beam may be required (to prevent weak axis buckling of the spandrel).

I agree with an axial demand possibly necessitating the deeper incoming beams / connections but disagree that it would have anything to do with weak axis buckling, at least in the literal sense. Lateral support anywhere on the beam -- and of any depth -- will restrain weak axis buckling. The only bracing benefit of the deeper incoming beams / connections is to restrain torsional buckling modes. Granted, many designers will take the computational shortcut of evaluating torsional buckling indirectly by instead assessing the lower capacity weak axis buckling between points of torsional restraint.

Quote (dik)

I almost never use full depth plates... usually stop 1" or so from the top flange... it's cheaper and easier to fabricate. In the project above connection was by clip angles.

That strategy may or may not be appropriate in this situation. Holding the stiffeners back from the flange makes sense primarily for dealing with concentrated loads applied parallel to the beam web. For a torsion situation, where the loads are perpendicular to the beam web, there is merit in grabbing the flanges directly, sans gap. For stability bracing, I'd usually be willing to hold the stiffeners back. For directly applied torsion of any serious magnitude, I would not.

Thanks... torsion with clip angles is minimal and easy to accommodate with the clip angle connection... bolt centroid and half the web tickness at most... generally include that in the connection to the purlin connection and 'wish away the torsional component'.

Rather than think climate change and the corona virus as science, think of it as the wrath of God. Feel any better?

-Dik

So at a spandrel beam we connect only from one side, and there is inherently an eccentricity in the connection from reaction load at bolt distance + 1/2 tw. but would that trigger using a full depth stiffener to "eliminate" the torsion? I cant imagine anyone is designing a typical spandrel W section for torsion in the general case.

Quote (sauce man)

So at a spandrel beam we connect only from one side, and there is inherently an eccentricity in the connection from reaction load at bolt distance + 1/2 tw.

In my experience, most engineers will call that a concentric connection while, intellectually, acknowledging that:

1) It surely does induce some torsion in the girder and;

2) The torsion is likely -- and substantially -- resolved by the connection transmitting torsion in the girder to flexure in the infill beams.

Quote (sauce man)

..but would that trigger using a full depth stiffener to "eliminate" the torsion?

No, it wouldn't. But there are other sources of torsion demand in a spandrel beam including:

3) The need for lateral torsional buckling resistance, particularly if you have negative flexure in the girders.

4) Direct, transverse loads applied to the bottom flange of the girder by the cladding system where the setup would induce such loading.

5) Torsional column buckling where your girders experience axial demand from their engagement as diaphragm chords and collectors.

Quote (I cant imagine anyone is designing a typical spandrel W section for torsion in the general case.)

I wouldn't normally...

Rather than think climate change and the corona virus as science, think of it as the wrath of God. Feel any better?

-Dik

Quote (sauceman)

So at a spandrel beam we connect only from one side, and there is inherently an eccentricity in the connection from reaction load at bolt distance + 1/2 tw. but would that trigger using a full depth stiffener to "eliminate" the torsion? I cant imagine anyone is designing a typical spandrel W section for torsion in the general case.

The moment in the connection doesn't really become torsion in the beam. AISC does have requirements on "rotational ductility" of pinned connections. The intent is that the connections should be flexible enough to be classified as a pinned connection.

I wasn't thinking about this in this thread. But, I'm guessing it could be an issue if the bolt line of the supported beams are pushed out farther than normal.