Marten:
We need to understand the detail to answer your question. Show us a sketch, with some dimensions, loading conditions, etc.
If it’s really a built-up section, then you do the flanges and webs separately. You can turn them, welding on alternate sides, to control the weld distortion. After the first few passes on one side you can roll the piece, back gouge the original root, if full pen. is req’rd. and make a few passes on the second side. Then go back to the first side for a few more passes. The web or the flange will end up flatter this way. You need run-out tabs at the two edges, but when you are done you can clean up the edges.
Then you fit the two flanges and the web together and do the fillet welds btwn. them. Do each sides of the bot. flg. to the web; then flip the beam and do both sides of the top flg. to the web, and again alternate t&b if multi pass fillets are req’rd. On large girders or heavy welds, I would weld on both sides of the web at the same time, again to control welding distortion; this will tend to keep the flange at 90° wrt the web. This t&b alternating will minimize the camber (weld distorion) in your final beam.
If they are all good welds, designed for the conditions and stresses they’ll see the exact location shouldn’t really make much difference. Obviously, you try not to make flange welds in max. moment regions, or web welds in max. shear areas.
I assume the original poster is asking about how far about the shop butt splices in the web plates and flange plates need to be. There isn't a formal requirement; D1.1 clause 5.20 specifically says you can put them all in the same plane, though most people think that's a bad idea. Our house specification calls for a 6" separation.
