CAESAR II has the functionality to consider load case combinations to perform fatigue analysis against a defined S-N curve. When considering fatigue due to different expansion stress cases this is fine. When considering pressure cycles there seems to be no capability in CAESAR, nor any clear guidance in B31.3. For piping specified as high pressure, B31.3 K304.8 requires a fatigue assessment in accordance ASME Section VIII Div 2 App5. This requires knowledge of peak stress intensity range which cannot be determined by CAESAR. For normal service piping reference is made to B31.3 301.10 which states that "cyclic loadings shall be considered in the design of piping", but provides no guidance as to how assessment of pressure cycles should be performed. Furthermore, the rules in B31.3 302.3.5(d) only allow for combinations of different displacement stress ranges and not the stress range due to pressure cycles.

I therefore seem to be left with the task of preparing a fatigue analysis and thereby determining a fatigue life utilisation factor for the pressure cycles externally to CAESAR, and then combining it with the that calculated for the expansion stress range from CAESAR manually. Is my interpretation correct?

Steve Baldwin

Actually, CAESAR II is equipped to consider pressure cycles in its fatigue analysis, in a limited way.

Fatigue analysis was introduced in Version 4.10 of CAESAR II in order to support TD/12, the gas transmission code for the UK. This code gives explicit instructions on how to evaluate cyclic loads (in terms of the stress calculation method, the stress concentration factors, and the fatigue curves to use), for both displacement and pressure loads. Since CAESAR II now had this capability for one code, we decided to make it available to users of non-TD/12 codes as well -- but since, as you point out, there are no guidelines in these codes, we had to come up with our own.

When the user designates a load case as having a FAT stress type, CAESAR II (1) calculates and reports the stress intensity (the same as required by Section VIII, Div 2), (2) uses the SIFs for the particular code in question (rather than peak stress concentration factors used by Section VIII, Div 2), and (3) compares the allowable to the fatigue curve entered by the user (for convenience, CAESAR II provides a number of reference fatigue curves for different materials). So you could set up load cases such as:

1 W+P1+T1+D1 (OPE)
2 W+P1 (SUS)
3 W (OPE)
4 DS1-DS2 (EXP)
5 DS1-DS3 (FAT)

or:

1 P1 (FAT)

Either one of these will provide a Fatigue analysis for a pressure only case.
The only thing incorrect about this result is that it will not include peak stress concentration factors (since the SIF for pressure under B31.3 is 1.0). To gauge the magnitude of this undercount vis-à-vis Section VIII, Div 2, we can look to Section III, Subsection NB (Nuclear Class I piping), which also requires fatigue analysis of piping be done to the Section VIII, Div 2 fatigue curve. This code requires that PD/2t be multiplied by the stress indices C1 and K1 to estimate the peak stresses – the product of these two indices ranges from 1.0 (for straight pipe) to 5.4 for socket welds to 6.0 for butt welding tees; butt welds range from 1.1 to 3.5 depending on the weld preparation.

My advice would be to set up your fatigue cases as described above, but use less than the entire Section VIII, Div 2 fatigue curve, dividing those values by the maximum Subsection NB C1 x K1 stress indices for your fittings (for example, 6.0, if you have tees present). You can do this either by (1) maintaining your stress ratio less than 16.7% in the Fatigue stress reports, (2) typing in a lower value when entering your fatigue allowables, or (3) changing the “Stress Multiplier” near the top of the .FAT file you wish to use by multiplying it by 6 (this will basically divide the data in the file by 6 when reading it in).


------------------
Tom Van Laan, PE
COADE, Inc.

One question,
In pressure cycles, you talk for socket Weld use 5.4, but in the table NB-3681(a) there is
Socket Weld C1= .... ; K1 = 3

So i ´d think the value c1*k1= 3?

How do you know C1*K1= 5.4?

Thanks in advance.