Sam,

This is a good question. Here is the response from Tony Paulin from PRG (for those who do not know PRG are developers of FEATools and other FEA software for PVP industry, www.paulin.com).

SIFs are a function of the basic geometric parameters d/D, D/T and t/T, and are a function of the component “type”. We can address the issue more specifically if you can either send or describe the input. Barrel type fittings have t/T problems that can be confusing. Reinforcing pads include additional parameters, one of which (the pad width), isn’t included in the B31.3 SIF.

B31.3 does not really provide a torsional SIF and so the FESIF or FEATools i-factor output for torsional SIFs can be greater than 1.0, the default for the Code, and in this case the B31.3 calculated stress can be too low by the value of the i-factor. This is considered the single largest omission from the B31.3 stress evaluation, especially when the D/T ratio is larger than 50. A rough estimate of the torsional SIF is (io)x(d/D).

The problem may be “perceived” only when the thickness of the branch is larger than the thickness of the run. In this case the B31.3 SIF is somewhat “hidden” in the effective section modulus where the Code requires the section modulus of the branch to be calculated by the section modulus of the run. The approach used in the Code can result in stress estimates that are too low when t/T << 1. The stress estimates are probably reasonable when t/T>1, the actual SIF however should be found from (i)(t/T). (See WRC 329 section 5.0).

FESIF also has an iK calculator that will compare correlation results that give an appropriate value for the B31 Codes for comparison.

Larger D/T junctions can also be sensitive to the weld thickness and the length of the attached pipe. FESif permits the input of both of these parameters that are not considered in the Code SIF.

If you would like to post the branch connection type and geometric properties, a more specific comparison can be made.

In short, if your system is going to cycle more than a few thousand times, or there will be some other loading that can potentially cause fatigue failures, then the proper SIF should be computed and used, and the FEA approach corrects a number of issues identified in the B31.3 code equations for SIFs in Appendix D.

Regards,
Tony

-------

If you want us to review your files, send in your CAESAR II file and FEATool input file to the Intergraph Smart Support at http://smartsupport.intergraph.com/