Hello,

I'm agree with CraigB that using the same SIF value used for 90 deg branch connection is incorrect.
But I think that is incorrect using the same SIF value used for 90 deg branch connection and the header.
I note that Caesar in simulation uses the same values.If we use formulas by B31.3, we note that the SIF value depends on pipe thickness.So, if we have 16" heather pipe and 8" branch pipe connection with different thickness, for my opinion the using of the same SIF value is incorrect.
B31.3 formulas are very conservative and could create several problem in overstress condition. Probably, using different SIF valus, we should have less problems in this kind of simulations.
Also in 45degree branch connection the analyst has the same problem, and there's another problem: the area of 45degree branch connection on header pipe is bigger then the area of 90degree branch connection on heather pipe. So, for my opinion,45 degree SIF value will be higher then 90degree SIF value.
Are you agree with my opinion?
But in what way we can find the real value of SIF 45degree branch connection? Could be a solution using a proportion beetwen areas of 45degree and 90degree branch connection?
I explain this concept: geometrically 45 degree branch connection area is 1.41 times bigger then 90 degree branch connection area.So the radius of 45 degree branch connection area is bigger then 90 degree branch connection area.The flexibility characteristic (h) changes.So it could be possible using the different flexibility characteristic value to obtain the real value of 45degree branch connection

Could be a solution using a proportion beetwen areas of 45degree and 90degree branch connection?
I explain this concept: geometrically 45 degree branch connection area is 1.41 times bigger then 90 degree branch connection area.So the radius of 45 degree branch connection area is bigger then 90 degree branch connection area.

No. If area (moment of inertia) is bigger than SIF must be smaller. FEM modeling confirm It.

When we analysed results of branch FEM shell modeling we discovered that SIF for 45 degree branch is smaller than for 90 degree branch.
But the stress concentration factor (peak stress) in real life is much greater in 45 degree connection than in 90 degree branch. It confirmed by experements. Look WRC 360 (by E.C. Rodabaugh) for example.
i=K*C
K - stress concentration factor. Much greater for 45 degree
C - peak stress from FEM modelling. Lesser for 45 degree
Inplane SIF increase very insignificantly for 45 degree (because moment of inertia is increasing too and C decresing), but outplane SIF increase significantly (moment of inertia become the same, but K is increasing).

The experiments sited in WRC 360 for SIFs all involve proprietary fittings and not pipe-to-pipe intersections. Rodabaugh in WRC 360 compares the work of Wordsworth, Smedley and Khan for pipe-to-pipe intersections, and these comparisons show the reduction in inplane and outplane SIF for laterals. (The SIF reduction is believed to occur because the lateral geometry provides a larger footprint on the header.) In fact, PRG NozzlePRO results compare more favorably to the Worsworth data than the Hsiao and Khan results given in the WRC 360: 8.7% to 17%.

Shell elements are the element of choice for most SIF work done for EPRI and by PRG for the ASME. Where D/T ratios become small, the shell solutions are supported by brick models, but the end result is typically a linearized membrane or bending stress which is then used in a fatigue evaluation.

As in the PRG article sited below in the note, the moment loading of most concern for laterals is often torsion since there is such little data for it, and since the torsional SIF can become larger than the inplane or outplane SIFs when then intersection angle becomes steep.

The "special coefficient" mentioned below is suggested when pressure cycling is the load mechanism of concern. High stresses due to pressure generally occur at locations away from the high stress locations due to external loads. This makes combining pressure and external load stresses using the method of stress intensification factors difficult.
Fortunately, many of the situations where pressure cycles are the major fatigue causing mechanism can be analyzed separately. We recommend an increased SCF for pressure cycling loadings. This is based on model comparisons with cyclic pressure failures - principally those of Deacock. (Ref 5 in WRC 360).

The complaint that there are no experimental FSRFs (or SCFs) for laterals goes somewhat against the variety of test data that supports full penetration welds of different types and the fact that the high stresses for out-plane loads do not occur in the acute crotch weld of concern.

The most significant cost associated with fatigue testing for PRG is the preparation of the specimen. If the lateral issue is important to this particular client, we would be glad to run Markl style fatigue tests at no cost on several laterals providing the specimens were prepared and shipped to our facility. We are also interested in any new SIF or fatigue data of piping components that may have escaped our recent literature search.

Regards,
Tony Paulin