crane girder overstressed

huangyhg

crane girder overstressed
heavy duty crane girder has been in operation since the late 40s.  numerous repairs have been performed at the support connections (mostly due to poor detailing-plastic hinging of the web and deck plate).  analysis indicates that the section is overstressed at the center span under documented loads and tolerances (picture shown with out-of-tolerance rail).  
q1 - since the original design allowed this overstress (approx 2 ksi or 10% of allowable for a7 steel) and the girder has considerable history of adequate performance what additional evidence or percautions are necessary?
q2 - the deck plate does not have sufficient connections with the girder to cover composite action.  won't the existing bolts shear due to overstress?  how would you fix this?
teguci-
how bad is the section overstressed?
i would only count on the "deck plate" to take side-thrust loading from the crane, not for composite action; especially in a "heavy duty" application or mill-duty crane.
the old aise tech report 13 covers these topics in detail.  
i would do a detailed visual inspection of the entire section and then do some nondestructive testing in selected areas (either liquid penetrant or magnetic particle), looking for cracks.  if it has been overstressed with cyclic loading, it could be approaching its fatigue limit as it has been in service for a long time.
you might also consider changing out the bolts in the connections or at least pulling a few of them and also doing nondestructive testing on them (wet method fluorescent mag particle would be best for the bolts).
stillerz,
under the standard 3/8" rail tolerance, i am getting a stress of 20.5 ksi for a7 steel (33 ksi yield).  the old code states a max stress of 20 ksi.  using current fatigue calcs the allowable is 21.8 ksi.  the drawings list an 18 ksi max (18 controls).  the major problem is the rails have moved and there is a possibility that the rail is offset up to 1.38" which gives a stress of 25.8 ksi.
even if the deck plate just sits on top of the girder, there will be a shearing action when loaded and the bolts connecting the two elements will fail unless they are connected to handle the shear.
ron,
i was thinking the same thing regarding the testing at the center span to check for cracking.  do you think a fatigue failure of the tension flange will be sudden?
the connectors are mostly rivets and some of them are gone.  for the ones that are missing, we will tell them to replace them with a490s slip critical and inspect every year.
thanks,
teguci...no, i would expect the failure to be ductile and progress from a propagation crack.
ron, are you sure?  i would think a fatigue failure starts with micro-cracking and that can be sudden/catastrophic failure in flexure starting with a brittle (strain-hardened) crack.  just thinking outloud with a keyboard.
i agree with ron's comments on the fatigue being the true limit state of concern.
i would not even glance at the "old" code provisions - they do not apply today - only perhaps offer you a way to understand what the orginal designer was thinking.  current codes would apply.
most of the problems with steel crane beams, however, usually start at the end connections with cracks in the steel around the upper end bolt (verses midspan).  
many times the combination of longitudinal rotation of the beam end due to loading in the span plus the lateral bending in the web (from lateral crane forces), just above the topmost bolt, initiates small, non-visible, horizontal cracks propogating from the bolt hole.
i'd second ron's suggestion for dye testing on the steel.

jae,
you are completely correct regarding the cracking at the connections.  the cracks in the webs have been repaired several times.  however, the conditions that cause the cracking in the girders at the connections can be addressed with better detailing.  the overstress at the center span would be a deal breaker.
as for using the old code vs new, the old code had a more conservative value for allowable stress than the new.  since we are looking at old steel (astm a7) i thought the original reference made sense.
that makes three votes for dye testing (including me).  always nice to have a sounding board.
thanks,

the old code is simply that...an old code.  despite the fact that the original was designed under it, the physical location of your crane now legally falls under the current adopted code and it should be followed. (standard of care, etc.)

teguci,
is the beam a rolled section as it appears in your sketch, or is it a built up beam?  if rolled, fatigue would be of less concern, and your calculations say it is satisfactory for fatigue, provided the rail were centred.
how is the rail fixed to the beam?  why has it moved?  how do you know whether it is the rail or beam that has moved?
agree with stillerz that the deck plate is only for bracing the beam, not for composite action.  for it to be composite, it would need to be welded.
it would help if you would describe the crane, its capacity, and its past and present use.
jae...once the cracking gets to a certain level, yes, the failure can be fairly fast; however, it will start with a small crack that progressively gets larger depending on the stress level. that's what produces the characteristic "beach marks" when you look at a fatigue fracture face.  
in looking at the sketch that i didn't see before, this is a fairly deep girder, so failure could happen fast, though i would not expect it to be catastrophic in the sense of the girder, though anytime a moving load drops it can certainly be a catastrophic event.
better to be safe and shut it down until testing is done.
hokie66,
beam is rolled (thankfully) and the allowable stresses have been computed accordingly.
rail clips are welded to the deck plate which in turn is bolted to the wide flange.  the rail clips hold the rail laterally.  not sure if the rail is shifted laterally relative to the girder.  a survey indicates that the rail is out of line at several points by 1.38".  i will have a look at this on site tomorrow.
unless the deck plate is allowed to slip longitudinally along the top of the girder (slotted holes i suppose), then it has to be connected based on vq/i.  if the two elements are connected firmly then why wouldn't they act compositely?  i agree that the buckling of the plate in compression needs to be checked (similar to what aise pub 13 shows), but ignoring it for composite action doesn't seem appropriate.