Hi,

There's flat, rectangular tubesheet, and half-round manifold, welded at the straight edges to the tubesheet. The semiround edges are closed with flat, semicircular heads.

Now let's calculate strength of the semiround manifold. Applicable equation is in UG-27(c)(1). It takes E to compute it. No problem, we have Table UW-12 with all those E's. It's a category C longitudinal joint, corner-type, listed as Type No. 7. Going to the right, we find value of... NA. Footnote (7) explains: "There is no joint efficiency E in the design formulas of this division for category C and D corner joints. When needed, a value of E not greater than 1.00 may be used".

Am I free to choose E here?

_________________
Konrad Anikiel

Hello Konrad,

You are right, there is no joint efficiency given in Table UW-12 for such a corner joint. Type no (3) gives an E of 0.6 for a single welded butt joint without backing strip and no RT. For lack of a better solution I would accept E = 0.6, provided your manifold is an exact halfpipe. This means a contact angle of 90° between halfpipe and tubesheet, consequently no moment acting on the weld seam.

We have calculated such a design several times and compared it with the results of an FE-calculation. Experience has shown that this will generally lead to a conservative result.

Michael

Michael,

I do not understand your reply, the corner joint is NOT a butt joint and consequently Type 3 is not applicable. It is Type 7, and the UG-27 calculation for circ. stress does not have a weld, therefore, you can use the E accpording to UW-12(d), 1,0 or 0.85 when there is a circ. butt joint.

The corner joint goes under UW-13, right?

Dirk

Michael: there always be bending, due to flat plate (even supported tubesheet is not infinitely stiff). Are you sure that you didn't set a "fixed" constraint to it in your FEA?

Dirk: right, UW-13 is applicable to the joint. But what is applicable to the semi-round shell? I need to calculate thickness of the manifold! I would agree that UW-12(d) applies, when it would be bolted to the tubesheet. But it's welded. It's quite unforeseen situation, when a longitudinal seam in a cyllindrical shell is of Category C. Or did I categorized it the wrong way?

PS. The joint in my case is like in Fig.UW-13.2(e-1). Austenitic stainless, TIG welded, no radiography, thin materials.

_________________
Konrad Anikiel

Dirk:

You are right, it´s no butt joint but a corner joint. Type no (3) has been mentioned as an example for a single welded seam without any RT.

However, the problem is not the circumferential seam but the longitudinal one. There is nothing in Table UW-12 which meets this specific design. If you want to stay on the non-FEA side you have to be as conservative as necessary.

Konrad:

Naturally the FE Calculation considers bending at the point of attachment between halfpipe and tubesheet. With UG-27 you can determine the thickness of your shell, however, without any consideration of the tubesheet attachment. My experience shows that taking a factor E = 0.6 will give a result which is conservative.

However, feel free to perform your own FEA and let us know what you have found out

Michael

Yes, I'm tilting toward deeming UG-27(c) as not applicable. It's not a cylindrical shell, and as considered here, cannot be treated as such.

PS There's not only FEA available. Personally, I don't treat it as the ever applicable and authoritative method. There's so many mistakes possible to do in the model... You can never be sure whether all assumptions and simplifications are right. In this case we'll use bursting test, because the tubesheet is burst-tested anyway.

_________________
Konrad Anikiel