Is this the correct way to get an SIF using FEA;

1. Model the header pipe, branch connection and weld detail
2. Apply a bending moment to the branch
3. Apply an equivalent bending moment to the pipe with no branch connection
4. Divide the stress with branch by stress without branch to get the SIF

This wouldn't account for geometrical imperfections so would you add a margin? The stress should stay proportional to the bending moment shouldn't it (My/I)?

To the extent I know, it is slightly different.The key element here is the use of the SCF (Stress Concentration
Factor) to get from the theoretical shell calculations to the tested fatigue
results. This is essentially the WRC 107 Kn and Kb, and is somewhat a
function of the element type selected, intersection weld model, etc.

Regards

Hello,

It is important that you not confuse stress intensification factors with flexibility factors. Loading a model with unit loads and developing a ratio of deflection (strain) to unit load will provide insight to flexibility but will not give you B31 SIF's.

I think the way you are describing will not calculate the "equivelent stress". Its not that easy.

I suggest the WRC (Welding Research Council) Bulletin 392 "Developing Stress Intensification Factors: I - Standardized Method For Developing Stress Intensification Factors For Piping Components", 10/30/2003, E.C. Rodabaugh ; G.E. Woods.

Look here:

http://www.forengineers.org/

And search for WRC Bulletins and use the "392" number for your search argument.

Also, do not confuse stress concentration factors with stress intensification factors. And remember that the B31.1/B31.3 SIF will be about one half of the ASME Section III (NB, NC type) stress indices. The B31 SIF's came from testing one company's, NPS 4, schedule 40, carbon steel fittings (and some fabricated branch connections) and they assumed that an appropriate SIF for a girth butt weld was 1.0 (?)!!

The testing placed three test samples in the fixture (one subjected to in-plane bending, one subjected to out-of-plane bending, and a straight piece of pipe with a girth butt weld in it) and it cycled them together until they failed (one at a time in fatigue). The SIF was just a ratio of the cycles to failure of the fitting to the cycles to failure of the piece with the girth butt weld.

In recent years Glynn Woods and E.C. Rodabaugh have been doing more testing and they are suggesting some chnges to the Code. Seems that the quality of the weldment is very important. Also, look into the "Schneider Effect" - the branch to run ratio has a lot to do with the cycles to failure.

Well, the good news is that the B31 Codes tell you to use better SIF data if you have it. Just make sure that you have SIF data per the WRC 392 method.

Regards, John.

Further to John's response, I would like to add few points.

Typically the commercial Finite element analysis programmes which shows SIF also shows SIF for torsional , axial, pressure loading etc.
The thing to remember here is that cyclic pressure for example, is not considered in the B31.3 code for pressure design, so when there is a strong cyclic pressure component, some intensification should be included. The rationale used here by these programmes ( to the best of my understanding )is the same as that used in NB3200. One must also remember that the "i" factor reported in Appendix 3 should be adjusted to get from the nominal stress in the branch to the peak stress (for example) by a factor equal to the effective section modulus multiplier. (See WRC 329). This means that the "real" stress intensification factor for branches is (i)(t/T), and this is what these programmes come out with - the "real" stress intensification factor.


Regards

Anindya Bhattacharya

ASME B31.4-2004 Para 301.10..."Fatiuge due to pressure cycling, etc. .... shall be considered see...."

B31.3 does not provide detailed rules per se but does provide some warning on pressure cycling.

Anindya is referencing WRC 329 and the titile is : WRC Bulletin 329 by E. C. Rodabaugh (entitled, "Accuracy of Stress Intensification Factors for Branch Connections") This publication is an excellent summary of the known problems of developing SIF's and suggests solutions to address those problems.

Regards, John

Thanks John and Anindya. I will order a copy of WRC329 to guide me on this problem.