Is this what you mean? Either way is OK.

• http://files.engineering.com/getfile.aspx?folder=34c93dca-d61f-4394-8e24-81

Regardless of code, it depends on the length of the beam, its section properties, the load and if the beam is braced laterally by any other framing.
That being said, I rarely see an issue with beams with unbraced tops in residential construction.

AELLC,

Yes, that's exactly what I mean. However, in this case the "header" isn't a door opening; it's the opening for a garage, two cars wide, with two floors and a roof bearing down on it. The "header" is a steel beam and there is a 2' stud wall above it. I'm not involved in the project but I know people who are. I would like to provide them with a quick clean reference on the issue if such a thing existed. My understanding is that the state of Virginia used to specifically disallow it in their code but perhaps I'm wrong. I can't seem to find it in there now.

Excel,

Yes, I agree. And this case seems to be well beyond something I would feel comfortable with. But, not being directly involved in the project there doesn't seem to be much I can do other than to suggest that the homeowner consult with the engineer of record, assuming there is one.

Archie264:
I don’t know if there is a prescriptive code requirement one way or another, but it is common sense engineering that you must pay attention to the fact that you might be introducing a hinge condition, and must design to account for that. There should be no problem taking vert. loads or lateral loads in the plane of the pony wall. And, that’s as far as most builders think of the problem. But, you have to take the lateral loads perpendicular to that wall another way. If we both have the same definition of pony wall, I’ve seen the problem more often as follows: a five or six course conc, blk. found. wall, with a 4' high stud wall on top of the blk. to the 1st. fl. framing. And, this produces a hinge at the mid height of the found. wall. Basically, the conc. blk. wall cantilevers from the footing to retrain soil pressure, or other lateral loads.

Usually the header (if in the low position) is designed for the vertical load and lateral load. Then the lateral load is transferred to the king posts, then to the diaphragm above/below.
I would also tend to agree with Excel that in general you don't see many issues either way.

I believe there was a post about which is the better detail a short while ago with some information. I'll see if I can find it.

EIT
www.HowToEngineer.com

For garage doors, you want the drop header so that there is capacity for wind bending normal to the header. However, a steel beam usually has much weaker Y-Y axis than its X-X axis, and with 2' of cripple wall above, there is some question about top flange stability against lateral-torsional. If you have the steel beam high, then the problem is bracing the top edge of the door opening against wind normal to the door (although the roll-up door doesn't transfer wind to the top edge).

Even with 2 floors and a roof, why can't a glulam or Parallam header calc out?

O.K.... given that AELLC’s sketches define the problem, the engineering logic would be the following: by putting the long steel beam right up under the fl. framing, you provide better lateral support for it (the primary load carrying member), from the fl. framing and diaphragm above. Then it feels a bit awkward to hang the 2' of cripple studs and a door header below the primary beam, but you still need some lateral strength/stiffness at the top of the door. If you sheath this entire detail with plywood or osb it should be just fine. When you put the steel beam right over the doors, you provide a good header with some lateral strength/stiffness, but the cripple stud/wood pl. on top of the steel beam does not provide good lateral bracing for the stl. bm. because of the of the hinge at the top of the beam. I’ll bet that was Virginia’s thinking on the matter. If you sheath this detail with plywd. or osb you end up with a plywd. box beam w.r.t.vert. loads, and a stressed skin panel resisting lateral loads and helping stabilize the stl. beam. The bigger problems, we now know, have to do with how this wall of the garage works as a shear wall, when 80% of it is door opening, and providing enough framing around the door opening and a strong enough door, so the door doesn’t just crumple inward in a strong wind.

Thanks folks.

AELLC, as for using a glulam or parallam, well, it's not my project and I'm not involved. Someone asked what I thought of the configuration, which isn't much, incidentally. He seems to no longer be concerned so I'm backing out of it, other than for purposes of professional knowledge of the situation. I sure wouldn't design it that way but I don't suppose whoever's project it is much cares what I think about it in any case.

Thanks again.