I asked Tony Paulin to comment. He has not review the current corrspondence. Here is his response:

To produce the piping related SIF from a finite element program requires the
following:

1) Compute the peak stress from the finite element analysis. Be careful thinking that increased meshes give peak stresses. Increased meshes around as welded geometries (effective notches) result in solution singularities and cannot be used. (See WRC 429 "3D Stress Criteria Guidelines for Application" and WRC 474 Master S-N Curve Method for Fatigue Evaluation of Welded Components" for FEA guidelines and recommendations.)

2) Be sure to use the restraint and loading definitions that are associated with a typical Markl-type fatigue test. See ASME B31J-2006 "Standard Method for the Determination of Stress Intensification Factors (i-Factors) for Piping Components by Test" (not sure if this has been released yet.) A good discussion can also be found in Nureg CR/3243 for SIF's, fatigue, tests, and the ASME Code rules.

3) Find the nominal stresses (M/Z), PD/2t, F/A, for the matching pipe and the corresponding load as discussed below by Dr. Becht.

4) Divide the range of peak stresses by the nominal stress caused by the same range of loads and then divide by 2, not letting the value become less than 1.0. The SIF thus derived can be used in a B31, beam-type analysis of a piping system. For reduced intersections, the user must also be sure that an effective section modulus is not used automatically by the pipe stress program with the user's defined SIF, otherwise the SIF must be further modified before use.

5) Many people use the actual loadings from a pipe stress program and perform a stress analysis per ASME Section VIII Division 2, Appendix 4 and 5 in accordance with B31.3 304.7.2 for "unlisted components". The actual combinations of loads are used since this is often simpler than generating SIF's. The singularity precautions mentioned above should still be observed when computing any peak stresses. The Master Curve methods removes this concern however.

I'm not sure these comments help, but they accurately address the issue as I understand it.