I am deeply confused and doubtful about all the fatigue assessments I did in the past with CII as I have recently done a little research on the fatigue calculation according to ASME VIII div.2.
I think CII calculates max 3d shear stress of every element in the fatigue scenario and compares it with the relevant ASME fatigue curves values.
my first question is: why do fatigue load cases consist of only the source of fatigue? (say in a system which is subject to cyclic temp. variation between T1 and T2, we choose T1-T2 and compare the stress results with twice the ASME values; shouldn't we go for W+P1+T1 and W+P1+T2 and compare the results with one time ASME values)
my second question: since the fatigue load case in ASME is Pb+Pl+Q+F, F being the peak stress, how could we calculate/include F in our CII assessment?

thanks in advance

Hi there,

Here is my personal opinion. Maybe it helps...

Pb+Pl+Q+F stress is already calculated by as B31 expansion stresses, because B31 SIFs are Peak Stress Intensification Factors...
...BUT, be careful, ASME B31 SIFs are defined with respect to girth butt weld, which includes an intrinsic SIF = 2.00.

Therefore, to calculate the overall Pb+Pl+Q+F ASME VIII-2 peak stresses with respect to a "smooth" non-welded spool, you need to increase twice the B31 SIFs.

Personally, I use to reduce two times the ASME VIII-2 fatigue stresses and to let SIFs unchanged in the model - it's faster, with less risk of making mistakes and it's enough accurate since the highest stresses occur generally in fittings and not is simple pipe spools.

As per ASME VIII-2, the Pb+Pl+Q+F stresses, when are checked against fatigue allowable, should initially be divided by 2 (meaning that actual Secondary + Peak resultant stress ranges, as calculated analytically or by FEA, are reduced twice, to find the stress amplitudes) and then are checked against Sa limit from S-N fatigue curves.

By default, Caesar II considers two times the ASME VIII-2 fatigue strength from S-N curves and therefore there is no need to divide the calculated stresses by 2. But, as I've written above, it remains the SIFs non-conformity issue...

Regarding the loading cases development, it is user/engineer option to consider the relevant cyclic loadings for fatigue check. Is it Pressure Load cyclic load with SAME NO. of loading-unloading cycles as Temperature increase/decrease? If pressure and temperature experience cyclic variations simultaneously, then yes, P+T should be considered in fatigue check.

BUT is it Weight (W) a cyclic load? Excepting fluid weight action, I don't think so...

Anyway, if T1 = Tmax and T2 = Tmin < Tinstall, then I don't think W+P1+T1 and W+P1+T2 are recommended for Fatigue check. You need indeed the thermal expansion/contraction RANGE T1-T2 to be checked. Then, you may need to calculate the EXCLUSIVE PRESSURE Stresses as (W+P1) - W.

Finally, it depends what requirements you have from Client/Owner and what assumptions and/or method statement have been agreed in the Design Basis...


Regards,

I am still having issues with why CAESAR II multiplies the stress values from the S-N curves by two?

Any additional clarification is appreciated.

Thinking out loud here, but perhaps they're not doubling the stress but halving the cycle.

When you go from state 1 to state 2, it's a half-cycle. Going back to state 1 would make it a full cycle.

Fatigue curves reflect stress amplitude. The piping code expansion stress range is just that, a range, or two times the amplitude.
So in static analysis input, CAESAR II will double the amplitude to give a range for these code-defined range checks. But if you go on to a harmonic analysis - which is an amplitude calculation - these curves will revert to amplitude by halving the entered stresses in the static analysis fatigue input.