Tank Rafter Lateral Deflection Limits while in Service

[steeletc]

First time poster, hoping this meets the forum criteria with respect to duplicate threads.

I just wanted to re-open the discussion on a closed thread in light of a recent client request:

https://www.eng-tips.com/viewthread.cfm?qid=420043

I believe the structure in question was a storage tank of sorts, where API 650 is typically invoked for the design in North America. For the design of roof rafters where the plate sits on top, API 650 permits the rafters to be considered laterally braced when loaded from the top under live or snow load, but specifies they must be treated as unbraced for dead load only as the friction is likely considerably less.

The OP for the referenced thread was looking for tolerances on the sweep of these rafters in service, which neither API 650 or API 653 provide direct guidance on.

In a similar situation now where 3D scanning data has been provided with a request to "assess" the vertical and lateral deflections. Only guidance available is in ASTM A6, which gives sweep tolerances of 1/8" * L/5 (with L in ft) for narrow W sections, and 1/8" * L/10 (with L in ft) for wider Ws. For channels, the tolerance is marked as "negotiable" which makes it difficult to draw a line in the sand. Even then, this is only a mill tolerance for fabrication and erection, so I can't see how this would be considered a reasonable value for "inspection" purposes.

I agree with the OP that the small deflections beyond a reasonable limit are likely locked in due to friction between the rafter and roof plate, and these could be knocked straight. The cause of the lateral deflections is likely some form of LTB, though even lateral deflections up to 4" may be within the elastic deformation capacity of the channels.

Has anyone since arrived at a reasonable limit for lateral deflections? I'd like to think L/200 or L/180 could be considered acceptable as some research has suggested that the ultimate bending resistance for elastic LTB is not "significantly" reduced (say, cut by more than 30%, for with LTB is only considered under dead loading only in accordance with API 650). Dead loads are small compared to snow loads. I'd also imagine that the stress in the beam from being bent laterally would need to be exceeded prior to additional lateral torsional movement because there would be a residual elastic stress in the member (i.e. the rafter out-of-straight is not truly an initial out-of-straight but rather a residual displacement).

I consider this a very gray area overall worth discussing and debating in further detail than offered in the original closed thread.

Maybe KootK is out there somewhere...

 

[JoshPlumSE]

That looks a little tricky, but I'm not sure it's all that crazy.

The left side frames out the same way as the left side of Sectin AA, right? The right side (right of the 2nd ridge beam, frames out similarly). It's really the transition between the two ridge beams that's the concern. Right?

I wonder if you can raise or lower one ridge so that the transition between them isn't noticeable from the outside?

What does the architect expect this to look like?

 

[JStephen]

One thing to add to the original post is that in the older API-650, the requirement about not considering friction with dead load only was not there, so the arrangement in question quite likely wouldn't meet the current standard.
It is also possible for rafters like that to flop sideways for part of their length, and deflection while remaining more or less vertical would be less of an issue.

 

[steeletc]

JoshPlumSE it's not an issue with the transition but rather an issue for one rafter in isolation. Consider the deflected shape of the rafter under lateral load as a simple beam: what's a reasonable limit on that deflection given that there's not really any lateral load present? Default would be ASTM A6 sweep tolerance but again, not truly applicable in my opinion and very vague with respect to channels.

JStephen this is a good point though as the snow/live loads often exceed dead loads by a few times, this doesn't appear to be an issue very often - there's usually plenty of capacity omitting friction even for the analysis of older roofs.