When performing a Fatigue analysis using Caesar using PD5500 Fatigue Curve F2 I have a query. Since the general ASME Fatigue curves are based on Alternating Stress vs Number of cycles and as Caesar calculates the Stress Range it is possible to compare the Stress Range values directly against the Fatigue curves from ASME. However as the fatigue curves in PD5500 are based on Stress Range vs Number of cycles and the Caesar Stress Ranges are nominally 1/2 the actual stress range does the input of the fatigue curves from PD5500 have to manipulated to ensure the Caesar calculated Stress Range (50% of the actual) is compared to a relevant Fatigue curve from PD5500?

Anyone willing to contribute/answer ?

If you build a fatigue file (.FAT) with a stress multiplier of 0.5, the data in that file should be stress amplitude. If you build a fatigue file (.FAT) with a stress multiplier of 1.0, or, if you enter your fatigue data directly in the CAESAR II static input, you should enter stress range data.
Confirm by reviewing the results.

Hi David,
I understand the process of inputting fatigue curves. My query is rather more fundamental. If we analyse a system to B31.3 then the SIF's used for components are half of the actual SIF's and this is well known. So if we have a thermal stress range between T1 and T2 calculated we know the actual stress range in the piping is nearer twice the result given by Caesar. If we enter the fatigue curve (based on stress range values) to establish the allowable number of cycles with the Caesar computed Stress Range, which is less than the actual, surely we get the incorrect result. As we are dealing with fatigue should we not be using the actual stress range rather than the Code stress range to enter the Fatigue curve or does Caesar take this into account internally?

The CAESAR II fatigue file (defining stress AMPLITUDE versus Cycles) will double the stress value if used in the thermal stress range case. In harmonic analysis, the value used will not be doubled as we are calculating an amplitude there.
If you wish to incorporate a stress concentration factor, that's up to you. I would agree that doubling the SIFs used in B31.3 would better approximate the stress to be used in a more detailed fatigue evaluation.
If I'mnot confusing enough, let me add that one way you can accommodate this SIF increase is to cut the stress allowed in half, instead. You can do this in the FAT file by doubling the stress multiplier in that file.
Give it a try.

Hi Dave,
So if I am interpreting your response correctly if we input a fatigue curve defined by Stress Range versus Cycles (as per a curve from PD5500) then we should either:-
a) Apply a stress multiplier of 2 in the FAT file and run with the Caesar (Code)SIF's
or
b) Double the SIF values and run with the inputted Stress Range vs Cycles fatigue curve with a stress multiplier of 1.0

in order to do a more detailed fatigue evaluation.

There are too many ways that my answer could be misinterpreted.
1) You say "Stress Range versus Cycles". The (ASME) curves we ship with CAESAR II are Stress Amplitude versus Cycles. You can do either because we also employ a multiplier in that file. That multiplier converts your local unit dimension to CAESAR II native dimension (e.g. MPa to psi). That maultiplier can also be used to move from amplitude to range (or range to amplitude). That multiplier may also include a factor of safety, or, in your case a stress "adjustment".
2) You say in a) "a stress multiplier of 2". Is this to address your condition (I don't know how to better put this) that the "real SIF" for a girth butt weld is 2, not 1? If so, then, yes, that doubling effect can be applied with this multiplier. BUT, I cannot say the number you type in is "2". Note that for US units, the FAT file contains the value "0.5" is used so in that case, a value of "2" would quadruple.
3)In b), I guess you could double each and every SIF but your would have to define all your welds too. In concept, I think it would be OK but in application it's sloppy and error prone.

I have to again emphasize - give it a try. You will see how your entered srtess data (range or amplitude) is applied in the static (or harmonic) analysis. Yes, it's confusing but you can quickly confirm how your entered data s used by looking at the output. If the output is off, change that multiplier in the FAT file.

For those of you whoa re still reading this, I show two FAT files below. Note how the "Stress Multiplier" is used.

Here's out FAT file for low strength steel:

* ASME SECTION VIII DIVISION 2 FATIGUE CURVE
* FIGURE 5-110.1
* DESIGN FATIGUE CURVES FOR CARBON, LOW ALLOY, SERIES 4XX,
* HIGH ALLOY AND HIGH TENSILE STEELS FOR TEMPERATURES NOT
* EXCEEDING 700 F
* FOR UTS <= 80 KSI
*
0.5000000 - STRESS MULTIPLIER (PSI); ALSO CONVERTS AMPLITUDE TO FULL RANGE
*
10 580000.0
100 205000.0
1000 83000.0
10000 38000.0
100000 20000.0
500000 13500.0
1000000 12500.0
0 0.0

And here's one I built to mimic Markl:

* using Sc=Sh=20ksi with f in Eqn 1a
* will give the following fatigue curve
* use this with multiple sources of strain ranges
*
1.0000000 - STRESS MULTIPLIER (PSI); these are ranges
*
10 36000.0
3000 36000.0
7000 30637.0
10000 28528.0
15000 26306.0
20000 24835.0
50000 20677.0
100000 18000.0
0 0.0
*

Hi Dave,
Perhaps I have mis-interpretted your previous response. And I agree the second approach of doubling the SIF's is a "sloppy" approach but I thought that was a way you were suggesting. Can you advise how you would perform a fatigue analysis using Caesar II adopting the approach of DNV-RP-D101 which proposes PD5500. In particular how would the curves of PD5500 need to be input to Caesar II as the PD5500 curves are based on Stress Range vs Cycles? How do we ensure that the "actual" Stress values (taking into account the correct SIF's not the Code SIF's) are compared with the correct fatigue curve?

Hi Dave,
Your response
"You say in a) "a stress multiplier of 2". Is this to address your condition (I don't know how to better put this) that the "real SIF" for a girth butt weld is 2, not 1? If so, then, yes, that doubling effect can be applied with this multiplier. BUT, I cannot say the number you type in is "2". Note that for US units, the FAT file contains the value "0.5" is used so in that case, a value of "2" would quadruple."

has me confused a little as you seem to be agreeing initially that a stress multiplier of 2 will address the shortfall in the SIF's for butt welds( and more importantly the components in a piping system) but then you say the number is "2".

If we have a fatigue curve which is in US units and is based on Stress Range values vs cycles what would you recommend the Stress multiplier should be to perform a "detailed" Fatigue analysis?

For a FAT file in US units that uses stress range (rather than amplitude) and I want to simply double all SIF's, I would enter 2 for my stress multiplier. Then I would run a simple test to convince myself that everything is working the way I want it to work.

Many Thanks Dave for response.

Hi Dave,
Since the Fatigue curves provided in Caesar are from ASME Div II which are based on Stress Amplitude versus Number of cycles and then a 0.5 Stress multiplier is applied then the Fatigue data within Caesar is a Stress Range versus Number of cycles. Am I right in assuming that Caesar calculates the stress intensities based on "Code Stress Ranges" using Code Stress intensification factors? If so then is it correct to compare these stress Intensities against the Stress Range versus Number of cycles fatigue data to establish the damage as the Code requires that "true" stress intensities be utilised.

Additionally to my query above I have another question on analysis for fatigue. Can you advise as to whether SIF's should be incorporated into the piping model at welded attachments such as shoes or trunnions as these are structural discontinuities where I beleive SIF's are justified.

Hope you have the time to respond to this and my last posting since I want to ensure the fatigue aspects are covered correctly.

Anyone from Intergraph willing to respond to above two posts?

In the static analysis, when one of our .FAT files is used, the amplitudes entered in teh .FAT are doubled and appear as ranges to go with the typical evaluation of expansion stress range.
On the calculated stress side, CAESAR II will calculate a stress intensity using Code-defined SIFs. There is not stress concentration factor (SCF) used for those local (peak) stresses.

Thanks for the partial answer. My real question is as above - Is Caesar approaching the fatigue assessment in the correct manner by using "Code" stress intensities rather than "true" stress intensities ?

I beleive the Code intent (ASME VIII Div 2) is that "true" stress intensities should be used. Any comment?

Final question to which I hope for a definative answer.

Can you confirm that it is incorrect to analyse a system to ASME B31.3 and perform a fatigue analysis using Caesar when the fatigue curves used are the Caesar standard ASME Curves with a Stress multiplier of 0.5 since Caesar is comparing "Code" calculated stress intensities rather than "True" stress intensities especially for bends and fittings?

Please advise of a better way to get direct responses to queries regarding Caesar applications.

I would not expect the fatigue evaluation to be valid when the polished bar (offered in the CAESAR II .FAT files) data is used without modification...unless I polish the pipe.

How about this from L.C. Peng's book - Pipe stress Engineering - page 445 "By doubling the i value, we have increased the [B31.3] stress to the same level of theoretical stress. ... However, since the i values of most common piping components are automatically calculated by most computer programs, we may use the original i and just double the calculated stress." But he goes on to say "The i method used by B31 ... assumes that every inch of piping consists of a girth butt weld without identifying the actual girth butt weld locations".

Thanks Dave,
First response is a little "light hearted" and I know we cannot use the "polished bar" fatigue curves directly - but do the majority of Caesar users and there is no warnings in the Caesar documentation about this??

Your second posting still does not give a definative answer to my last question but skirts around the issue. As generally the high stress points are at bends/tees etc it is at these locations which are not checked correctly in my view by Caesar if you use the caesar .FAT files. Caesar compares these "Code" stresses (well intensities) which are not "True" stresses, which you acknowledge, and compares against the fatigue curve to establish a damage value. Therefore Caesar will underestimate the damage by not using the "True" stresses in the bends/tees etc.

Can you answer directly my previous posting by either confirming or disagreeing with the statement?

DSB1954UK,

Finally, what have you concluded?

Did you get further response from Dave?

Good questions.
But what is the final answer?

Hi PKU & Sanok7,

In my opinion, Dave's latest response in clear enough for this issue.

As per my understanding, Coade/Intergraph will never provide design & analysis solutions outside of Codes' applicability. Therefore, Dave's considerations may be regarded as guidelines/recommendations for fatigue assessment of B31 piping.

Please keep in your mind that B31 Codes and DNV RP D101 just refer to ASME VIII-2 Part 5 and/or BS PD5500 Annex G in relation with fatigue assessment, but do not provide "recipes" or specific methodologies for further action.

At the time of his post, Dave's considerations brought to me the latest confirmation regarding some doubts I used to have based upon Paulin Group (PRG) webinars related to fatigue analysis.
So, in my opinion, this approach is the most suitable one to be employed for B31 piping systems' fatigue assessment.

Of course, there are many other "key" issues the engineer should take into account for when he/she performs fatigue analysis on B31 piping piping systems, such as: WHAT KIND OF STRESSES are to be checked against fatigue Curve? or What Fatigue Curve to use?

The first question is very important in my opinion. I believe that Dave's approach is mostly applicable for ASME VIII-2 Part 5 "Smooth Bar" Fatigue Curve. I've seen companies performing fatigue assessment of B31 piping systems using Caesar II peak stress ranges and PD 5500 fatigue curves (e.g. "D", "E", "F" etc.).
In my opinion, such approach is not quite accurate, since PD 5500 Annex G fatigue assessment methodology is not based on Peak stresses but rather on secondary stresses (as per ASME VIII-2 classification). Using "de-rated" PD5500 fatigue curves, the local/peak stress concentration effects are already considered with respect to weld type, weld & loading relative orientation etc.

A good option, many times suggested by our distinguished moderators (Dave and Richard) is to used IGE TD 12 Code within Caesar II environment for fatigue assessment purpose. If you'd have a look on IGE TD 12 fatigue analysis methodology, you'll note that IGE-TD12 SIFs are higher (around 2 times) than the typical B31 SIFs...and Pressure load SIFs are also provided, which you'll not find generally in common piping codes...

It just remains the engineer to employ proper judgement and sound correlations to perform a suitable fatigue assessment.

These are just my thoughts. Any other contribution/clarification would be welcome.

Regards,

Dorin Daniel Popescu, thanks a lot!