stress classification and FSRF for thermal stratification

[MBlackman]

I'd like to hear your opinion on classification of repeated thermal stratification for fatigue assessment using the smooth bar curve.
Hot/cold liquid stratification, or a half-filled horizontal cylinder can cause both thermal bowing and localized thermal stress at the level where an abrupt temperature transition occurs.
Bowing generates bending stresses which to me are treated like any other secondary stress for fatigue assessment (FSRF selected based on weld quality / NDE).
The local stress - should it be treated as a bona-fide local thermal stress under ASME VIII-2 , given the definition in 5-C.2.2 'no significant distortion'. In this context I would consider the bowed shape as irrelevant.
This clause also gives examples of local thermal stress which don't exactly fit this case.
If this is considered a local thermal stress then the corresponding FSRF is 1, whether for a seam or girth weld.
Or should it be considered a secondary shear stress and FSRF's apply?
cheers
Martin

 

[TGS4]

Can you please describe the exact equation that would cause you to not apply the appropriate FSRF to local thermal stresses?

 

[MBlackman]

I am talking of Sect 5.5.3.2 STEP 4, eq. 5.30 and the note (a) below.

 

[TGS4]

The FSRF should always be applied. Plus, per sub-paragraph (a), the likelihood that you are able to explicitly model both the notch effect of the weld AND the quality effect is essentially zero.

There are "issues" with the current version in VIII-2, which will not be fixed until the 2021 Edition. In the meantime, I recommend following the approach in API 579-1/ASME FFS-1.

 

[MBlackman]

Thanks very much for your response TGS4.
I will treat it as a principal stress (not shear) and apply the FSRF.
I note that API 579-1 2016 also has the option of separating out local thermal stress and not applying the FSRF (ref. Table 14-4), but it appears not to be the appropriate stress classification for this condition. If using FEA then 'local thermal' may be appropriate for peak stresses remaining after obtaining a linearized stress range along a classification line.