existing wood trusses are failing

huangyhg

existing wood trusses are failing...
i have any interesting project...
it is a 1-story plaza in hernando county, fl
it has wood roof trusses spanning ~50' between load-bearing cmu walls
it was built in 1985 and from discussions with the owner the roof originally was constructed of plywood sheathing with clay roof tiles...
upon inspection the roof has 1 member on each truss that has buckled, typical on every truss
the owner had originally contracted a "handyman" to provide "repairs" although no engineer specified the repairs.
members where added randomly as shown in attached photos.
i have since been contracted and modeled the roof truss with all applied loads to current code...
it appears that the trusses have no horizontal bracing
my questions:
the only member that fails (on-site) does not fail upon analysis... but other members do, why?
bad wood grade on every truss?
load transfer?
lack of proper permanent bracing?
current wind loads to excessive?
i have modeled the wood grades to be the minimum that will work for the minimum loads applied and not fail any members... but i don't feel comfortable assuming that the trusses where originally designed to not have any lateral bracing (rat-runs)
any thoughts on this process...
any suggestions or recommendations?

are you modelling the trusses as purely pinned?  they aren't... timber truss plates do transfer minor moment loads, which can have a significant effect on the load patterns...
what size are the plates?  i have a 1980s text on the capacity of the plates.  perhaps modelling with a modified spring stiffness or nodal moment loads will give you more realistic load patterns.  as you've already seen, our models don't always match reality.  the designs we produce often work more because of the general rule that if a total system has sufficient capacity it will redistribute as needed in service to achieve stability.
very interesting situation...  keen to see how it all works out!
cheers,
ys
b.eng (carleton)
working in new zealand, thinking of my snow covered home...
have any localized distributed or point loads been added to the trusses - either on the roof or hanging from the ceiling underneath?
structuraljoe....those trusses were probably locally produced (lakeland, tampa, brooksville, dade city, etc.) so you can probably find the original truss drawings, even if the building department doesn't have them.  many of the larger lumber companies have small truss manufacturing operations tied to them.
the trusses were probably designed for bottom chord bracing and it was likely shown on the truss plans, but the contractor might have left them out.
if i'm understanding your photo, the failed member failed in compression.  further, it looks like either a gambrel roof or a mansard.  is that correct?  if so, both pick up large lateral wind loads.
i agree with ys....our models don't always equate to reality, and the most common error is that of connections not acting like they're modeled.  do as ys suggested and see how your distribution changes.  don't necessarily look for failure in that member, but look at its axial load relative to the other   
"the only member that fails (on-site) does not fail upon analysis... but other members do, why?"
can you post a sketch of the truss, identify the loading mechanism, the failed and the survived   
if my memory serves me correctly, i believe that the northwest received some bad lumber from canada around that time.  i don't recall the particulars, but there was a big flap about it.  you might look into the grades provided.
mike mccann
mmc engineering
almost every roof truss failure i hear about during heavy wind loading happens because the web bracing was inadequate or not installed.
  
mdj
never ask "why" with wood.  i've seen so many things that should have failed but didn't.
i've seen a rectangular shaped house have one of its perimeter basement walls completly collapse leaving the house to "span" the long distance from one side to the other.
wood structures have so much crazy redundancy that they don't always follow traditional stability logic.
thanks everyone for all of the input,
attached is a jpg of the truss along with failed members from the program i am analyzing it in...
all i can determine is that the "other" members are overstressed but due to some of the modifications present (from the "handyman"... that just cracks me up every time i type it... i can hear it now "yeehaw i sur can fix yo probem... i'll nail me some 2x8 here and there and you's be good to go...) it isn't as noticeable on-site.
as to the cause i would think that even if this structure had experienced full dl and maybe full ll or wl i doubt it has ever seen current design load combinations dl + ll + wl (nds - lfrd)...
i intend to add the lateral bracing and reinforce some of the members that fail in my model and go from there...
does anyone know of any procedure to determine the grade or strength of the wood in the truss?
i have designed the modeled the truss with current loads applied and based the lumber grade on the effectiveness of the truss... but i have no way of knowing if the truss was properly designed in the first place and if it was truely designed for a clay tile roof... the owner says it was originally a clay tile roof but that wouldn't be the first time i've been lied too...

that looks like ramadvanse