PD5500 high cycle fatigue analysis for use on a piping system.

[Winnower]

Hi, is anyone familiar with using PD5500 for high cycle fatigue analysis in piping system according to DNV-RP-D101?

I am running various load cases using CAESAR II to identify the different stress ranges in my piping according to the rain flow method as required but:

CAESAR II outputs two stress values when you run the "STRESSES EXTENDED REPORT":

1)the "Max Stress Intensity" (column 6)
2)the B31.3 "Code Stress" (column 9)

My question is simply, which of these two stresses from the "stresses extended report" should we be using as input to our PD5500 high cycle fatigue analysis?

I believe that we should be using the "Max Stress Intensity" multiplied by a suitable factor for our PD5500 high cycle fatigue analysis because it takes account of the hoop-pressure stress (amongst others) which is important for fatigue analysis as one of the principal stresses.

In my opinion the B31.3 "code stress" (sustained stress + expansion stress, B31.3 para 320 and 319) should not be used because it does not take account of the hoop-pressure stress and only takes account of the longitudinal pressure stress (amongst others) because the hoop stress is accounted for in the B31.3 code in para 304.

Does anyone have any opinions either way on the above argument?

regards,
Winnower

 

[DriveMeNuts]

Neither of the stresses that you list.
It is the normal stress range perpendicular to the expected crack.
For a crack running parallel to a longitudinal pipe weld, it is the circumferential principal M+B stress range at the surface which is used in the analysis.The Stress Intensity will produce a correct result for this example.
If you want to assess a crack parallel with a circ weld seam, then using the stress intensity will produce a conservative result.
I don't remember there being a high cycle adjustment factor. The S-N curve takes care of that.

 

[Winnower]

Thanks for your reply DriveMeNuts,

I guess we don't know what the orientation of the crack will be so a multi orientational, oblique, "3D" approach would probably be desirable. Isn't that what the CAESAR II max stress intensity gives us?

My understanding is that fatigue cracks initiate at a mircoscopic level when dislocations occur within the crystalline structure within the grain. These dislocations (plastic deformations) are due to an oblique stress coinciding with the grain structure yielding the material. A fatigue crack occurs when successive dislocations occur causing a fracture at the grain boundary. In my opinion, since the grains within a structure can be orientated in any direction (I think?) it follows that the crystalline structure within the grain can be oriented obliquely so I think I need to consider stresses acting in an oblique direction.
Once the fatigue crack has initiated at the grain boundary, then it believe it could run along a longitudinal or circumferential weld, but in my opinion the initial dislocations and initiation of the crack could well be in an oblique plane/direction which is, I think, what using the "max stress intensity" gives us, the maximum stress in an oblique direction relative to the X, Y and Z axes of CAESAR II.

As for the "high cycle adjustment factor", I think that a correction is required purely due to the way CAESAR II works, it delivers the stress amplitude and not the stress range. I have not got as far as considering this factor fully yet.

My main query was whether the "Max Stress Intensity" output from CAESAR II would be the appropriate stress to use in high cycle fatigue analysis being a "real stress" based on a 3D stress state. Whereas the B31.3 "Code stress" is a partial stress which does not consider the hoop stress.

Thanks again for your reply, it's got me thinking,

regards,
Winnower.

 

[DriveMeNuts]

Your description of what happens at the microscopic level sounds about right.
However, PD5500 deals with the macro level, where as the micro crack grows, the resulting 'visible' macro crack orientates perpendicular to the direct stress. This approach has been found to be successful through testing for many simple geometries.
I think Using Stress Intensity will work with geometries specified in PD500 because it simply funds the greatest direct stress at any location, but it will likely be conservative.
For more complex stress geometries not specified in PD5500, the ASME method may be better as it uses linearised Von Mises which considers stress in all directions.
For super crazy stress fields, no fatigue method works, and laboratory testing is the only option. There is an appendix in ASMEVIII D2 with details of how to do this.

 

[MJCronin]

Post in the "COADE, Inc: CAESAR II" forum ....

MJCronin
Sr. Process Engineer

 

[BJI]

What is cycling? If the displacement stress range above 100,000 cycles why not follow Appendix W?

 

[Winnower]

Thanks all for your responses,

BJI, I'm doing high cycle fatigue for piping systems on an FPSO, so it's the piping that is cycling and we are considering the hogging and sagging of the hull, the operating pressure cycling around the process value, the temperature cycling around the process value and slug loads amongst others. As for appendix W, the clients preference was for us to use PD5500, so that is why we are using it.

MJCronin, I posted a similar thread in the COADE/CAESAR forum but it didn't gain any traction and when I looked at PD5500 queries, unsurprisingly, most of them were in this forum so I tried my luck here and have had some useful discussions.

DriveMeNuts, thanks, yes I considered using CAESAR II's own high cycle fatigue analysis module which is based on ASME III and ASME VIII fatigue guidelines but as I mentioned our client preferred that we use the PD5500 as is mentioned in DNV-RP-D101.

What I believe is that whichever high cycle fatigue failure theory you use, hoop stress is part of the calculation of the principal stresses S1, S2 and S3 (ref Mohr's circle). From these stresses you can calculate the octahedral shear stress to implement Von Mises theory or the maximum shear stress to implement Tresca theory.

As I understand it CAESAR II can output either the maximum shear stress (Tresca) or the octahedral shear stress (Von Mises).
If you use CAESAR II's own high cycle fatigue module, it bases it's analysis on the maximum shear stress (Tresca theory) using the "Maximum Stress Intensity" output. What I am trying to establish with this thread is that the other stress output from CAESAR II from the "Stresses Extended" report is the B31.3 "code stress", and because that does not consider the hoop stress, it is not appropriate to use the B31.3 "code stress" for any of the high cycle fatigue theories.

regards,
Winnower.