Hello,
The B31 Codes were based on experiments done with standard welding fittings and with welded (fabricated) fittings. The Appendix "D" SIF's for elbows reflect tests on 90 degree elbows. The Code recognizes the limitations of the information in Appendix "D" and says that if you have a better SIF (more accurate for the fitting at issue) you should use it. It is a take it or leave it sort of thing. B31 Appendices "D" are not perfect but they are what we have for now. It is appropriate to use this data until something better comes along.
We should remember that the intensification of the calculated (beam theory) stresses is a function of the degree of ovalization of the pipe (elbow) cross section as the elbow deflects. As the ovalization increases with deflection, the flexibility becomes greater and the effective stresses increase proportionally. Also, the effect of the ovalization DOES NOT stop at the weld line (although we routinely turn on the SIF and FF at one weld line and turn it off at the other weld line). The pipe will ovalize with deflection beyond the weld line and into the adjoining straight pipe. As a 45 degree elbow deflects, more of the ovalization will be carried beyond the weld line (relative to a 90 degree elbow). And as we all know, flanges on the ends of the elbow will affect the ovalization (and the flexibility and stresses) as will other factors e.g., internal pressure and the wall thickness (most of the B31 tests were done on NPS 4, schedule 40, seamless carbon steel pipe and fittings).
There are many factors like this that will confound the piping engineer's attempts to accurately model a piping system for structural analysis. Think about this when you are fretting about how much you should "round-up" your spring hanger loads or what the SIF should be (within 3 decimal places) for a 37.255 degree elbow. If you come within 10 percent of the actual sustained loadings you were lucky. The B31 Codes have proven to normally have enough design margin (factor of safety or index of ignorance) built into them to assure safe piping systems. There are after all practical limits on how much accuracy we can achieve in modeling systems.
As for a reference, see Dave's posting regarding the NOW AVAILABLE (for a mere $35) Tube Turns Piping Engineering book. Remember this book includes the Markl papers and the experiments alluded to above were done by Markl at Tube Turns. Some additional work was done later (at Oak Ridge by Sam Moore and others) on larger diameter pipe (with several bend lengths) and the resulting publications can be found in TID 25553 (if you can find that tome).
TID--25553; SURVEY REPORT ON STRUCTURAL DESIGN OF PIPING SYSTEMS AND COMPONENTS. Gathered and edited by Rodabaugh, E.C. ; Pickett, A.G. 1970 Jan 01
http://www.osti.gov/energycitations/searchresults.jsp?Author=Pickett,+A.G.
Regards, John.
_________________________
John Breen