Hello all,

I am doing the stress calculation of a horizontal header with 36" Sch.10 that it has a short vent branch 8" Sch.10s in the top of the pipe section of the header with reinforcing pad.

In the first approach we have modeled that branch directly from the center of the header and we have inputed the SIF as reinforcing pad in the intersection between header and branch. The result obtained for expansion case was 90% of allowable.

How can be that possible? This branch with a free end in expansion case should have almost free expansion with no stresses and the stress value obtained in the intersection should be just the stress value of the header pipe (13% approx.)

We will change the model in order to avoid this result. We will model a rigid from the centerline of the header pipe to the wall and we will add a FEA SIF in the node of the branch element. But.., I would like to understand the 90% stresses in expansion case obtained from the first approach.

Than you very much in advance

Regards

Maybe it's the effective section modulus if I'm not mistaken. But if it's a free end something is not right since there's is no restraint. But if the configuration of the header in both sides this branch is not flexible it is the header driving this high stress at the free end branch not the branch driving this high stress.

Modeling rigid element and using FEA SIF is a good approach since actual section modulus will be used. The D/T for 36" S10 is 115.4 so this a good candidate for using SIF's derived from FEA.

Just correct me if my interpretation of your system is correct.

Cheers,

Some correction of my statement below:

But if the configuration of the header in both sides "OF" this branch is not flexible it is the header driving this high stress at the free end branch not the branch driving this high stress.


Cheers,

But based from your statement the header stress is only 13%. So it's either the branch effective section modulus and SIF from Appendix D of ASME B31.3 that drives this high stress.

Any correction is welcome.

Cheers!!!

First of all thank you for your answer.

I agree with you. Expansion stresses in the header pipe are only 13% of allowable. After modeling the branch with FEA, thermal stresses are 0% in the branch node as I expected since the free end results in a free expansion of that segment and the stresses in the pipe header remain the same 13%.

Thank you very much for your help

Regards

You're welcome. In cases like this you need also to check in FEA SIF for the header. I have made a run for header SIF and it's essentially 1 (I have found 1.19 for in-plane header SIF which is very close to 1). Just double check this in your case.

Any other opinion is highly appreciated.

Cheers!!!

I agree with your result for header SIF Thank you very much.

Regards

B31J has equations for BOTH the RUN and BRANCH SIFs. This was a long time coming. The run versus branch SIF issue has been known for a long time but you needed to do an FEA or physical test to quantify the header SIFs which are typically very low for d/D<0.4. Below is a snipet from a Paulin Research Group (PRG) Webinar042 from 2008:

Refer to WRC 329 (E. C. Rodabaugh 1987) Para 4.4 Run Moments:

"To illustrate a large inaccuracy in all (B31) Code i-factor equations ... we give the following example ... UFT (28"x0.375" and 2"x0.250") ... it is intuitively apparent the ... i-factor (for this example is off by 10)."

"In discussing this kind of example with users of B31.3 and B31.1 we have been told, in effect, that the Code requirements are obviously silly; the piping analyst should use his judgment and ignore the vent or drain line in his ... analysis."

Tony Paulin had been preaching this and other Code branch connection shortcomings long before I first met Tony in 1993. Tony's persistence and passion led to the B31J and of course making FEA more accessible to the average engineer.

Well said Bob. I've been watching many webinars also from Mr. Tony Paulin. I've learned a lot from them. The degree of confidence of engineer will become high in designing reliable piping systems (such as those systems subjected to cyclic loading). But of course, the best way is to have a solid foundation on where this concepts come from, not just inputting and reading output from software. As we know, FEA is an approximation technique. Therefore, it is still up to the engineer to validate his results like doing convergence study.

Any other opinion is highly appreciated.

Cheers!!!