Thoughts - Glulam truss not recovering from snow load deflection

[swsengineer]

Just wondering if anyone had some ideas that I haven't thought of.

I have a relatively new (3 years old) glulam truss that spans 75' over a Firehouse. The roof experienced an above average snow load in December and now the truss is deflected 3.5" at centerline and is very noticeable from the outside in the metal roofing/tectum panels above it. It seems the deflection occurred from the snow but for some reason the truss has not responded and come back to it's normal dead load only deflection. Dead load only deflection should only be about 1" I and the original engineer agree with this. The truss members do not appear to have split or been damaged in any way. It was suggested that reaming of bolt holes to facilitate erection is to blame, but their are lots of bolts and even with several web members removed in the truss model the dead load deflection only changes slightly. The deflection by everyones account was not present before the snow load and previous although less snow loads in the past have not presented a problem.

Another thought was the 3/8" side plates may have been put on too tight with the nuts being too tight instead of a snug fit. Friction in the plates or some kind of clamping forces is not letting the truss members rotate back into place.

Possibly something in the tectum roof panels shifted and something is preventing the truss from coming back.

The truss is on the hip end of a roof with a flat bottom chord, partial sloped top chord, then flat top chord for the center 3/4 of the span give or take.

I didn't engineer the truss but I don't see any issues with the design.

3.5" over 75' isn't too bad but the problem is this truss is 8' away from a cmu wall so 0" deflection in the roof 8' feet away make the 3.5" deflection jump out and be very noticeable.

I realize I'm not giving enough info to present the full picture. I'll be glad to give more info to specific questions. Really I just wanted to see if anyone had some good ideas on what might be happening.

Thanks,

 

[woodman88]

What are the current deflections of the other trusses? It may be that all the trusses are deflecting about the same but the truss nearest the end wall appears the worst.
If the glulam members appear to be intact, I would question the joints. The deflection (of 3.5" in 75' span) could be caused by the top chord joints compressing and the bottom chord joints slipping.

Garth Dreger PE
AZ Phoenix area

 

[swsengineer]

other trusses have less deflection around 1" but they are also a different configuration.

bottom chord is one piece 75', top chord is in 3 pieces, 2 sloped ends and continuous middle flat chord.

 

[woodman88]

I would first look at the gussets at the bearings to see if I could determine if the gussets have rotated or the lumber has slipped. For a single bottom chord 75' long to deflect 3.5” at the center the bearings (or the bottom chord to move in from the bearing or top chord) only have to move a total of a 1/32”. Do the other trusses also have a single length bottom chord?
What sizes are the top and bottom chords of each truss? Are the trusses all 8' o.c. or do they have difference spacings between them? And are you trying to determine if the truss is structurally okay or will you be trying to "fix" the truss?

Garth Dreger PE
AZ Phoenix area

 

[swsengineer]

yes the other trusses have single length bottom chords. The tributary loading of the other trusses are similar. The chords are 5" x 12 3/8" with 5"x6 7/8" web members.

This particular truss is more highly stressed than the others under full DL+LL but still ok, barely. I work for a glulam manufacturer and design trusses like this all the time. I would have designed the truss to be a little bigger not due to deflection but due to stress. Having said that the truss has more stress that I like but nothing unreasonable. We design most of what we sell but in this particular case the design came in from the project engineer already sized with connection details. Our engineering dept never saw it as it went straight into production from the project engineers sealed drawings.

Now 3 years later they have a problem with the one truss and are trying to figure it out and have asked for our opinion. They suspect whoever installed it over reamed holes to make putting it together possible. Only problem with this to me is the deflection from DL would of occurred right away not 3 years later. And the connection joints have 20-30 bolts so you'd have to have some serious field modifications to get these joints to have much slip. We prefit the trusses together in the plant and drill holes using the plates as templates. Also, usually any time an erector has to make any major field mod they backcharge and call to get approval. Nothing like that occurred. Anything is possible though. The trusses were shipped unassembled.

The thought right now is to shore up the truss, loosen the nuts and plate on one side and inspect the joints. re-install the plates..not too tight and If all looks good jack the truss up and see how it responds once the jacks are lowered. Will it go back to a 3.5" deflection or just have the roughly 1" deflection that it supposedly used to have before the snow load.

 

[woodman88]

Well if they only over reamed some of the holes than the dead load may not have been enough to deflect the truss the 3.5”. When the high snow load came it cause crushing of the not over reamed holes until all the bolts were working so you now have the 3.5” deflection. If the glulam member is okay at the bolt holes under the gussets, you might consider cutting the bottom chord and splicing it to shorten the length by 1/32” or so to take the deflection out. The big problem with this is that it is somewhat a guessing game to get it right or close.

Garth Dreger PE
AZ Phoenix area

 

[msquared48]

If the truss was wet or in high humidity before being loaded, then dried out after the load was applied, it could retain the deflected shape - same principle as a wicker chair.

I have seen this happen with glulam roof beams where the owner, to keep his pipes from freezing during the construction, fired up his wood stove and got a permanent negative camber in his wood roof from a heavy snow load too. Really worked well with his tile roof.

Could be the previous things too, or a combination.

Mike McCann
MMC Engineering
Motto: KISS
Motivation: Don't ask

 

[efsinc]

"different configuration".....
i am supposing you measured the unacceptable 3.5" deflection at the top chord as this is what would be apparent "8' away from a cmu wall" and therefore be objectionable.

have you measure deflection at the bottom chord?

 

[swsengineer]

the deflection was measured by a survey crew along the bottom chord relative to the ends and to the other trusses. This particular truss is essentially a parallel chord truss with only the last 7'-8' being sloped. The truss is fairly shallow with only a 5' centerline to centerline top to bottom chord depth. It has several vertical posts and webs so I doubt the top chord deflection is much different and the truss model doesn't show much difference.

 

[dhengr]

Swsengineer:

It’s pretty tough to comment in much detail without a sketch of the truss along with dimensions and loading, and without a plot of the deflections at each panel point. Are the deflections symmetrical and consistent (rate of change) from panel point to panel point? I don’t think this is an elastic truss deflection problem the way your program normal calcs. that. I doubt that the roofing system has much to do with the problem, it transmits gravity loads (DL & LL) but is not stiff enough in the plane of the truss to hold it in a deflected position. And, that system is loose enough in its own plane not to be imparting a significant tension or compression force in that plane.

With that truss under a roof plane change, I’ll bet you got more snow load than you think from drifting off the higher roof. And, you say the normal design stresses are higher than you would normally like to see. You say removing a few web members in your truss model doesn’t change the DL deflection much. But what percentage of the normal TL is the DL (abt. 25%), and did you remove the correct web members (allow the right movement at the right joint or joints) to cause the 3.5" deflection. You might test each joint or member with your truss model, if you can. Also check the bolt design at each joint for higher bearing stresses plus bolt bending stresses.

I think woodman88 has it right, he just didn’t go far enough. He talks about 1/32" movement at the bearings allowing the curvature and center deflection, and I didn’t check his math, but don’t doubt it either. I would look at all the joints, as he suggested, it doesn’t take much movement at several joints to cause the t&b chords to deflect, and they are certainly flexible enough, once a couple joints move a bit. Whatever way you drilled the bolt holes, slightly oversized or reamed, some of the bolts will be crushing wood before the remainder of the bolts come into play, and you said higher stresses than you would like to see. That might effectively allow a member to change length by 1/8" ±. Obviously, the tension diags. can lengthen from bolt bearing and bending alone. Whereas compression members may not shorten much due to bolt actions, if you had real good fit-up and sufficient bearing area on the sides or ends of each member. Tight side plates might have limited this movement under DL alone, and hold the joint, in the new position, after the fact. But, I wouldn’t expect much return of deflection once the joint is really seated even if the plates were loosened a bit, because this is not an elastic truss deflection issue. Also, I don’t really think there is much member rotation at the joints.

I would use a stretched piano wire or a laser and check the deflection at each t&b chord panel points, looking for uniformity and symmetry. Then I’d shore the truss, but not lift it yet, and inspect the joints. The repair will depend upon what you find during your inspection. Get the original engineer and the contractor who erected the trusses more deeply involved, they have as much or more at stake here than you do, assuming your plant did a good fab. job.

Good Luck

 

[efsinc]

this Is a puzzle.

the thoughts in my mind keep coming back to a) the fact that the truss of concern is a unique configuration in the roof structure, and therefore perhaps, field confusion led to the field misplacement of factory supplied webs and consequent bolt hole crushing hidden behind the panel point plates, and b) something collapsing / crushing in the tectum panel system above the top chord.

but, the fact that deflections at the bottom chord are 3 1/2" seems to rule these out. For my 'a' or 'b' to be relevant, the deflection as measured at the exterior roof line would be 3 1/2" and the bottom chord something less than that, say your calculated 1", pointing to a partial collapse.

 

[swsengineer]

Supposedly they are going to shore the truss up and remove side plates to inspect the joints. Hopefully that will tell us alot about what has happened. I will post the findings at that time. Thanks for all the input ..continue to share your thoughts.

also, the original erection crew the G.C. used was 3 guys in a pickup truck and the G.C. is no longer in business.

 

[HouseBoy]

How does the 3 1/2" deflection compare to the "predicted" deflection based on your best estimation of the actual loading that occurred?

How does the "above average" snow load compare to the original design case?

You write that 3 1/2" "isn't too bad". Is that in the generic sense or compared to the calculated deflection?

 

[swsengineer]

The predicted deflection for design snow load(20 psf) is 2.5". I'm not sure what value the "above average snow" load was, this was just according to the project engineer. To get 3.5" of predicted deflection the snow load would have to be around 37 psf. The truss is going to be thoroughly inspected in the near future and I am supposed to get an update when it does.

Thanks,