A bend should be considered "flanged" if there is any heavy/rigid body within 2 diameters of the bend that will significantly restrict the bends ability to ovalize.

Do forum members consider a tee or another bend within 2 PDs a heavy/rigid body that will restrict the ability of a bend to ovalize?

Ken,

To quote you, "A bend should be considered "flanged" if there is any heavy/rigid body within 2 diameters of the bend that will significantly restrict the bends ability to ovalize."

I'm not familiar with this rule, can you please cite your reference? It's the "within 2 diameters of the bend" that I haven't seen before.

Please advise.

Thanks,

B31.3-2004, Tbl. D300, Note 13.

Ken,

Over the last few years I have been able to check or review the work of several stress analysts from other offices. Surprisingly, I found that very few use Caesar's flanged-ell feature at all. The reason was they either didn't know that ells are stiffened by the flange or did not know how to take the additional stiffening into account.

Now, I personally always use CAESAR II's flanged-ell feature where appropriate. I have, however, done some experiments with it in the past and found that it makes very little difference to the results unless the ell in question is highly stressed with, or without, the additional stiffening. I'm not suggesting that the ell is not significantly stiffened--it certainly is; only that it makes little difference to real-world results.

But to return to your question, the answer is definitely No. The difference in stiffness between an ell attached to straight pipe as opposed to one attached to a tee or another ell is negligable and may even be lower.

Regards,

Ken quotes Table D but that's a 2D reference for tees. Markl's original paper states that 2 diameters of straight pipe are necessary to provide the geometry that allows the calculated bend flexibility.

I like how Ricardo thinks - he ran several jobs to develop his own (modeling) opinions and experience. And he's nice enough to share his conclusions.

But I would add that any "stiffener" within 2D of the bend would have an effect. And if I'm coming tight into an elbow from a branch, the header may stiffen the elbow. But again it may be a minor effect.

for the most part i only use that feature when modeling a trunion.

So the bottom line is, there is no 2 diameter rule in B31.3 for bends and B31.3 does not direct bends to be treated as flanged when flanges or stiffeners are simply close to the bend but not attached to the bend.

I can find no direct statement from the Code (nor do I expect such) regarding adjustment to bend flexibility and strength.

Well, my reference should have been to the Caesar Help screen for the "Type" field, which I believe is based on Table D300, Note 5, not Note 13. Thanks Dave more pointing out the incorrect reference.

Thanks everyone for your input.

The gist is: Some piping is modeled well by the beam element approach and the use of SIFs applied within the assumptions of code (Markl & WRC work), but Mr/Ms client this elbow is in a gray area, we need to go beyond what a beam element approach can tell us (e.g. FEA) if we need a better answer.

The 2 diameters stiffening is in Caesar User Manual section 12-6 quoting BS806 recommendations. With a bend trunnion clearly it cannot ovalise so the flex factor is 1 as for pipe. Technically the SIF is then unity but in fact there are local stress concentrations. So in this sense the situation is along the lines of a TEE fitting. But the TEE SIF would not be right since there is no hole in the bend. A trunnion on straight pipe also must have some SIF and here we know definitely there is flex factor 1. Current UK practice is to use "Kellogg's Method" for the local stress calculation, so it dates back to the big stress book of the 1950's.

"The 2 diameters..." Understood
"With a bend trunnion.." ok so far
"Technically the sif is then unity..."

From whoever's derivation you look to for bends Von Karman, Beskin, Vignesse and Pardue. The bend flexibility is due to it's ovalising and the stress concentration comes from the shell stresses associated with this. Stick a trunnion on it and it cannot ovalise. In this case the flex factor is one as for straight pipe but also those associated shell stresses will not be there. I surmise Winkler's curved beam theory should apply. I once calculated the SIF for this and it was around 1.03. Of course the trunnion is taking load so there is a local load effect with a SIF since the trunnion physically connects with only a part of the bend cross section. I presume the Hankin paper referred to in the Caesar manual correctly gives the SIF for this which should be quite localised. I suppose the bend could be split in two by the trunnion intersection and assumed to be stiffened in each part by an imaginary flange so allowing restricted ovality and accompanying flexibility and associated shell stresses elsewhere in the two parts of the bend.

Michael