ension field action question

huangyhg

tension field action question
hello all.  i have a question about plate girder "end panel" shear strength using tfa per the aisc spec.  the aisc spec. states that tfa cannot be used in "end panels" which are presumably the portion of the web between the bearing stiffener and the first shear stiffener.  
the reasoning listed in the commentary and in s&j 5th, bottom of page 553 is that the top flange (truss oriented like s&j5's fig. 11.8.2) must be present in the adjacent panel to provide horiz force equilibrium for the tension field resultant's horizontal component.  this is illustrated in s&j5's fig. 11.8.5--the little fbd showing forces at the top of the stiffener.  i do not understand this at all and my reasoning follows.
s&j5's fig. 11.8.2 shows a plate girder in elevation view, illustrating how it behaves like a regular square-end truss.  it is trusses 101 that the end diagonal for fig. 11.8.2's analogous truss is fully capable of being in tension because its horizontal component is resisted by axial force in the top chord *in the same panel*.  so why is the end panel of a plate girder not able to develop a tension field?
also, it seems to me that fig. 11.8.5 is wrong because the fbd of the top of the stiffener only includes forces from pure shear.  this seems like drawing the analogous truss node fbd and leaving out the force in the top chord to the left.
if someone can enlighten me, i'd appreciate it.
i haven't seen reasoning other than that stated in s&j, so i assume the provision is based on testing.  it makes sense to me, though.  if you look at s&j's figure 11.8.3, the web has buckled in the direction perpendicular to the tfa.  as such, it can no longer offer lateral support (in the vertical direction) to the flange, not a problem unless the flange is carrying compression from the tfa.  in a square ended truss, that   
thank you very much.
paraphasing to make sure i really understand the idea:
the flange can't provide a tf anchor by going into compression because the flange will just buckle in the vertical plane (buckle *about* a horizontal axis) due to lack of support from the buckled web.
correct?
yes, that's what i meant.  keep in mind, i haven't read this anywhere, and i'm certainly no expert.  it's simply how i see it.
i believe that you are correct and i appreciate your answer.  i know an ultra-guru on the subject so i'll ask him also and then report back.
a coworker of mine has "design of steel structures" by gaylord, 3rd edition. on page 485 they addressed this issue by stating that aisc didn't want to codify a method to handle the end panel tfas.
thanks eddy.
i just heard from an uber-guru and he said the answer is something different htan we've assumed.
he said that the tension field is very wide, so must be anchored over a significant depth of the girder, not just at the top flange.  at the end panel, there's nothing to provide anchorage over most of the height of the tf.