Dear All

Can you brief me about How to get(or define) the number of cycles for the fatigue analysis.I am aware of that it is not a normal plant cycle.I have not yet done any fatigue analysis. I am doing stress analysis for bridge piping.I understand from the design basis it is given that fatigue anlysis to be covered. Please enlightem me.

regards

Sha.

Your analysis method should be common to that used for the platform structural design, and the structural design premise ought to provide most of the background data you need. Look at DNV rules or similar and fatigue analysis using Miner's rule. The method sums the predicted life under cyclic action for a number of wave blocks with associated relative platform displacements and hence pipe stresses. Then using a suitable S-N curve a 'partial life' is predicted for each wave block. The final result is representative of predicted life in years.

Read paragraph 302.3.5 in B31.3. If this does not answer all your questions, read it again. If you still have doubts, either read it until you have no more, or find some other way to make a living. There's really nothing more about the subject than is contained in that section of the Code, which should not surprise anyne who understands that the B31.3 Code is mainly about fatigue analysis.

CraigB,

I partly agree with you ... however I suggest that when comparing the fatigue analysis method in B31.3, a code written over the years for onshore refinery and process plant, it is less appropriate to offshore wave induced cyclic conditions than the methods in DnV Guidance. DnV docs were written specifically for the arduous conditions encountered primarily in North Sea conditions.

B31.3 extended the cyclic factor from 10^6 to 10^8 cycles in order to address the high cycles associated with floating, production, storage, and offloading platform (FPSO) piping.

But there are are now doubts to this extention. Markl himself limited these f values to 10^6. Recent work done by Paulin Research Group (PRG)determined that Markl's curve is too flat and use at higher cycles may be unconservative. The paper reviewing the data was presented at last month's PVP conference in Chicago. PRG presented that paper in a webinar just yesterday. Perhaps you can contact them and review their presentation.

But this conversation is off the mark. The initial query was how to collect the number of cycles. I believe the people who do it right develop a Weibull distribution for these "random" loads, break them into groups and count from there. It's a guess backed by statistics.