Hokie’s terminology for the joints is a good way to describe this. If the moments on the jnt. tend to increase the angle btwn. the two members (the jnt. is opening) and the web forces or stiffener forces will be in tension. That’s the top jnt. in BA’s sketch from his post 29MAR11, 18:56. And, that’s a2mfk’s sketch also, which was exactly what I was trying to describe, except I would use only one stiffener on each side, right at the miter line. If the moments on the jnt. tend to decrease the angle btwn. the two members (the jnt. is closing) and the web forces or stiffener forces will be in compression. They might cause web buckling which must be accounted for. That’s the lower jnt. in BA’s sketch. The flg. forces, basically (flg. area)(the avg. flg. stress) = T, in BA’s sketch and are concentrated at that sharp transition and act akin to a point load, F, and cause a heck of a tensile stress concentration or causing web crippling just as a heavy, narrow point load on a top flg. of a beam must have stiffeners under it to prevent web crippling.
My comment about a curve as opposed to the sharp transition had to do with the fact that I saw this problem often on curved members with a radius to the inner flange “r” and a radius to the outer flange of “r + member depth”. And, the normal weld btwn. the flg. and the web which carries VQ/I shear flow, now also has a radial stress component perpendicular to the flg., and the weld must take both of these, or stiffeners must be used. This radial force component either pushes the flgs. into the web (BA’s lower jnt.) or pulls the flgs. away from the web (BA’s upper jnt.). This same phenomenon happens at the haunch in a laminated wood, three hinged arch, and shows itself as higher horiz. shear stress and a radial component perpendicular to the lams, and that radial stress must be limited or it’ll pull the lams apart, in tension perpendicular to the grain.
