questions regarding 121 span girder truss

huangyhg

questions regarding 121' span girder truss
i am working on the preliminary design of a fabricated steel girder truss (pratt) for a roof application.  the truss depth is limited to 9' by architectural constraints (unless i can show that it will not work). i have modeled the truss geometry in staad to size the members and to determine deflections.
i come up with a snow load deflection of 2" and a total load deflection of 4".  these deflections seem excessive to me, especially when compared to absolute deflection limits suggested in aisc design guide 3, serviceability design considerations for low rise buildings.  however, the deflections are less than typical limitations based on the span.  
the girder will be at the peak of a church roof, thus there are no ponding concerns.  there are no partitions that reach near the girder, but there will be a suspended gyp. ceiling that i am sure will require some special detailing for movements.  i am looking for some suggestions regarding acceptable absolute deflections for a long span such as this.  also, if anyone could direct me to a good design guide or text for special considerations that apply to long span trusses, it will be appreciated.
the design loadings and preliminary sizings are indicated below:
span:  121'
joist point loads:   3.0 kips dl
                     4.2 kips sl
                     5'-4" o.c. at panel points
top and bottom chords:  w10
vertical webs:  w10 with rigid connections to chords
diagonals: double angle
check out our whitepaper library.
it is a tad shallow for the span and loads but 'designable', none the less.  your ll deflection of 2" is good (l/720) with total load deflection of 4" (l/360).  there are no rigid attachments to be crushed from deflection and deflection can be accommodated by detailing.  if it deflects a foot, it still won't affect the structure and as long as the details accommodate the deflecton it should work.
from a design view, the deflection is small enough that there should be no second order considerations.
you might want to camber the system by 4" - 6" (an upward bow looks much better than a sag, although deflection may not even visable)
sometimes absolute deflection requirements are unreasonable or unworkable.  can you imagine a large suspension bridge limited to a 2" deflection?  deflection must be accommodated sometimes.
it might be prudent to inform the architect in writing of the anticipated deflection to ensure that his detailing relects this amount of movement.
dik - that was a great answer - totally agree with all you said.
this reminded me of a recent article i read about a bridge that is planned to connect mainland italy with sicily.  it has a proposed 3300m (2 mile) main span.  it would be interesting to know the anticipated max. deflection at midspan.  
let's see.....l / 360 for a 2 mile bridge is ... 30 feet!!
i hadn't thought of it in that perspective... no reason that it can't be designed for that amount, though...
i had a little 'experience' on a project a couple of years back with a pre-eng metal building... the frames were at 30' centres and they designed the girts for l/90; unfortunately the lower girt was supporting the top of an 8' high cmu wall... the girt on the shopdrawing looked a little flimsy and i asked the question... one of the reasons i like to review the shopdrawings for my projects rather than have another engineer do it.  i couldn't modify the girt myself without voiding the pre-eng building warranty and the 'extra to contract' that the owner paid for was several thousand dollars.  fortunately he could understand that the top of the wall couldn't accommodate a 4" deflection... moreover, he couldn't understand why they would design a building that deflected that much.  he had also been directly involved in the selection of the pre-eng building.
jae,
for your information, the deflections in bridges are generally limited to l/800 or l/1000 depending on pedestrian use, so the conventional deflection limit would "only" be 12 to 13 feet ;)
curvbridger - wow...12 feet makes me feel so much better!
the defelection values for the truss is not noticable compared to the span.
thanks for the insights about long span deflections.  revisiting the aisc design guide i notice that the absolute deflection limit of 3/8" to 1" falls under the heading "design considerations relative to interior partitions and ceilings."  partitions are not a concern here and ceilings can be detailed to handle the deflection.  
i would be uncomfortable designing a system like this for the minimum code limitations (l/240 sl and l/180 for dl+sl), but based on your comments i am comfortable with deflections of l/720 sl and l/360 sl+dl in this situation.  
dik,
you mention adding camber to the system.  is that commonly done with trusses?  my understanding was that it is not.  it seems like it would be difficult to obtain identical camber on top and bottom chords. if the cambers were not identical, then fit up problems with the web   
cambering throws a whole new wrench into the detailing geometry for trusses and there is a slight premium to be paid (with competitive bidding, however, it's usually minimal).  for custom, long span trusses, i have often spec'd a camber.   i generally camber top and bottom chords the same.  for long span trusses, the lengths of the web members may vary by a tad, but the detailler does the sums.
my comments about using a greater camber than anticipated is good for exposed trusses, it's best that they do not appear to sag.  all it takes is for someone to line the bottom chord up with a horizontal band on the far wall <g>.  for the same reason i often batter retaining walls into the soil (after a few years it doesn't look like they're falling over) and it's only a matter of tilting the formwork (simple and little or no added cost).
if the truss is completely concealed, there's no advantage in cambering it other than roof drainage.  i don't like ponds to reduce ac loads <g>.  the architect should be notified (in writing) that the total deflection can be a few inches and that he should provide sufficient joints, etc. to accommodate this movement (re  
gadit,
for a truss with a span/depth ratio such as you describe, initial camber is a good idea, and worth the effort.  a 3" initial camber gives you 1" up under dl and 1" down under dl + sl, which seem like nice, tolerable numbers.  
one other thing to note, that your gyp ceiling will be installed (presumably) after most of the other dead load has been applied.  so the sag for consieration there would be the roof live/snow loading and the gyp weight (and lights and other misc) but your mech, fire & structural dead loads should already be there before the gyp ceiling is hung.