Div. 2, fatigue analysis and exemption per 5.5.2 1

[jtseng123]

Dear all,

A vessel is exempted from fatigue analysis per Div.2, para 5.5.2.2, however, it’s cycles is greater than 10^6. Shall the fatigue analysis is required per 5.5.2.1 (c) ?

We are buying a new air filter duplicated from the past one which has been operating for more than 20 years without problem and without any fatigue analysis. It is a small vessel with legs supported and re-pad will be used to exactly duplicated form the old one. Only pressure fluctuation between 0 -100 psig in minutes at ambient temperature that the total cycles will exceed 10^6. We got stuck whether we shall spending money asking vendor to do fatigue analysis per 5.5.2.1 (c) even vendor stated they have never done any fatigue analysis or designed for cyclic service on such equipment and the equipment will be fine. This is an US vendor.

Can any one help to interpret the code between the two paragraphs ? Thanks.

 

[TGS4]

5.5.2.1(c) is very clear. You are not permitted to use 5.5.2.2 if the specified number of cycles is greater than 10^6.

 

[TGS4]

There is also something that you wrote that struck me after after I posted my original reply. You said

It is a small vessel with legs supported and re-pad will be used to exactly duplicated form the old one.

I take it that this refers to a fillet-welded reinforcing pad? You should note that even in 5.5.2.2, there are subparagraphs (a) through (f) that states that features such as those listed - your case comes in subparagraph (a) - that says that the possible harmful effects of the following design features shall be evaluated.

"Shall be evaluated" means that you still have to do a fatigue analysis that includes those features. So, even if your number of cycles was less than 10^6, with those features, you are required to perform an analysis, even if the component otherwise passes 5.5.2.2.

I should also ask whether you are going to be setting the thickness based on the current VIII-2 allowable stresses? If so, then there is no physically possible way that you could have an identical vessel - the new VIII-2 only came out in 2007!

 

[jtseng123]

TGS4,
This air filter is Div. 1 vessel. To make it identical to the old one is certainly possible with the same material, same thickness, same nozzle re-pad design with the same size of fillet welds even the latest code allows for higher allowable stress. It does meet 5.5.2.2 based on decades of service.

 

[TGS4]

Nope. 5.5.2.1(c) is very clear. You can't take advantage of the exemption.

And without going through VIII-2, you would have a very difficult time complying with UG-22(e).

Speaking of decades of service - are you aware of the design margin on (average) life built in to the unwelded S-N curves? It's 20. So, if you can demonstrate 400 years of successful service, then I might accept a 20 year life based on past successful service. otherwise you are degrading the design margin, which is contrary to U-2(g).

 

[Paulettea]

If the number of cycles is not more than 10^6 the exemption criteria ca be used.

"Shall be evaluated" means that you still have to do a fatigue analysis that includes those features. So, even if your number of cycles was less than 10^6, with those features, you are required to perform an analysis, even if the component otherwise passes 5.5.2.2.

TGS4, what do you mean by "do a fatigue analysis" here? Do you mean that if there are the mentioned harmful design features (e.g. fillet welds) we have to perform a detailed fatigue design or can we use screening criteria method A or method B?

TGS4, the code says a "successful experience over a sufficient time frame". what is a sufficient time frame? suppose that there is a vessel which is operating for 20 years and I want to fabricate the same vessel for 30 years then is it acceptable that I use this vessel as a successful experience over a sufficient time frame.

moreover, I need to know what is meant by the word "successful" in this context. suppose that the vessel operated for 30 years and no failure occurred. But in the 15th year is service there has been a crack detected and some repair has been done on it. can we still consider it to be a successful experience. What if there are many cracks developed on the vessel but fracture mechanics methods show that there is no need for repair. Is it a successful experience or not?

 

[TGS4]

Paulettea - to answer your first question, to perform an analysis means that an analysis per 5.5.3, 5.5.4, and/or 5.5.5 must be performed. In other words, skip over the Screening.

Regarding successful experience over a sufficient time frame, I shared my person thoughts on the matter in my 25 Oct 17 21:54 post. In your first example, using 20 years of experience to justify a 30 year life is simply ridiculous. Fatigue is a cycle-based, and hence time-based failure mechanism. If I was made aware of such an attempt to use 5.5.2.2, I would do everything in my power (likely very little, but nevertheless) to stop that abuse.

Regarding what constitutes success in the context of successful experience, I would turn it back to understanding the failure mode. In your example of 30 years of experience but a crack developed in year 15, I would not consider that to be successful - in fact, by the very definition of the fatigue failure mode, this failed in year 15.

Furthermore, if a fracture mechanics evaluation is required to demonstrate that cracks are self-arresting, then a fracture mechanics evaluation is necessary to demonstrate that cracks are self-arresting. I would only accept that the fracture mechanics assessment form a part of the vessel design book, not that a dubious claim be made under 5.5.2.2.

I would also add that I have also seen process licensors try to say that they have 400+ years of experience because they have 40 vessels that have been in service for 10 years each. No, that's not 400 years of experience, that's 10 years of experience 40 times over. Furthermore, my experience has been that most process licensors do not have adequate follow-up with the operating companies to know whether or not actual failures (cracks, etc) have occurred in service.

 

[jtseng123]

Back to my original post and I would like anyone who has in depth knowledge can provide technical inside why Dev. 2, 5.5.2.1 (c) is mandatory:

A typical vessel design life is 20 years (some contract may ask for 30 years, but that is not the concern here), that is what we are asking, not 400 years.

Vendor has built "many" filters (normally size is small)for decades with the same conditions and with all "harmful" features you can think of, such as fillet weld, re-pad, leg support (due to small size and for maintenance purpose), etc., and for many end users to operate for more than 20 years without a problem. It completely satisfies 5.5.2.2. There are reasons why it has no problem. One simple reason I can think of is its alternating stress is very low on nozzles, leg, etc since the operating pressure is cycling between 0-100 psig.

We are buying the same "harmful features" that have been proved good for 20 years or more based on vendor's experience. What is the technical reason we have to spend extra money asking vendor to perform fatigue analysis per 5.5.2.1 (c) for cycles > 10^6 ?

Code shall consider the real case and change the wording on 5.5.2.1(c), or offer technical inside why 5.5.2.1 (c) is mandatory regardless successful experience.

 

[TGS4]

5.5.2.1(c) has been in place (previously in AD-160) since the original VIII-2 was published in 1968.

If you truly think that the 10^6 limitation on the exemption is excessive, then please submit an inquiry to the Code Committee, along with your rationale. I note that this 10^6 exemption is not specifically mentioned in the original Criteria Document, and therefore may be empirically-based.

I take exception to your comment, though

Code shall consider the real case and change the wording on 5.5.2.1(c), or offer technical inside why 5.5.2.1 (c) is mandatory regardless successful experience.

The Code, or more appropriately the Code Committee shall do no such thing. They are not required to provide a justification for the rules implemented - only that the peer review process is sufficiently robust that 50+ engineers experienced in pressure vessels agree with such rule. However, as a member of the pressure vessel community, you have the opportunity - some might even say the responsibility - to submit your inquiry to the Committee(s), and probably attend the meetings to plead your case. Whether you like it or not, you are part of the ASME pressure vessel community and should take a role in shaping it. Complaining bitterly on eng-tips.com does not accomplish that. Do something!

 

[jtseng123]

TGS4, thanks.

 

[jtseng123]

TGS4, we have heavily debated internally whether 5.5.2.1 (c) is mandatory or not. AI from supplier said it is not applicable because it is stamped for Div. 1, and with sufficient experience per 5.5.2.2, the equipment is good to go.

 

[TGS4]

In my opinion, your AI is incorrect. Apparently the Code Committee needs to tighten such a loophole.

 

[Paulettea]

TGS4 I think there is a problem here. If the total number of cycles is less than 10^6 and there are the harmful effects mentioned in 5.5.2.2 there is no limit on using the exemption methods 5.5.2.3 and/or 5.5.2.4. My interpretation from your post is that when there is one of these harmful effects we have to perform a detailed fatigue analysis based on 5.5.3, 5.5.4, or 5.5.5.

Furthermore, I want to know what does the number of design cycles represent in the S-N curves in Annex 3-F. Is this the number of cycles to develop a crack or is this the number of cycles to fracture. I do not think that when you perform fatigue evaluation and fabricate a component based on the fatigue design there will be no crack initiation or crack propagation. It simply means that there will be no catastrophic failure.

 

[TGS4]

Pauleteeta - you are certainly asking all of the right questions.

Regarding what does failure mean in the context of the VIII-2 fatigue analysis, especially using the smooth-bar curves, it is not entirely clear. And that lack of clarity is nothing inherent in the method, but rather in the data that the curves are based on. Comprising the data is a mix of crack initiation (including a variety of criteria, including: 0.2mm, 0.5mm, and 1.0mm), all the way to Markle data which was leaking cracks. Suffice it to say that the Code Committees have argued this topic long and hard - to no conclusion. Furthermore, the curves generally represent a lower limit to the data, but have a margin of 2 on stress and/or 20 on life, based on the MEAN of the data. And remember that the data mostly comes from smooth, polished laboratory specimens, not as-built pressure vessel components. (Which provides a nice segue into my next topic...)

This is why I have always advocated that the fatigue analysis be used in initial construction to size components in order to generally achieve a desired life. That doesn't mean that you are then absolved from inspection throughout the life of the vessel. Rather, I strongly advocate on using a fracture mechanics approach (using either the method in VIII-3, or API 579-1/ASME FFS-1. Then, you postulate a hypothetical crack, just at the threshold of detectability, and then calculate the number if cycles to a FAD failure or a through-thickness leaking crack. Then, you set your inspection interval (based on the inspection technique and its threshold of detectability) at a fraction of that life. This approach doesn't guarantee any particular life, only that you will detect flaws before they become critical. (Note that the nuclear and subsea industries struggle with this, because they generally cannot inject in-service).

Regarding your first point, about using 5.5.2.2 when the total number if cycles is less than 10^6: the harmful effects of these less-then-ideal fatigue performance details are would carried by methods A & B. And 5.5.2.2 specifically says that those effects must be evaluated. Just a slight detour regarding wording in the Code... When it says that something must be analysed, then the engineer must produce and analysis and results. When it says that something must be evaluated, then some description of calculation must be produced, but it does not necessarily need to be an analysis (such as FEA). If it says that something must be considered, then the engineer is permitted and encouraged to apply engineering judgement that may or may not include an evaluation or an analysis. Make sense?

Method A has extremely low cycle counts, so is feverishly not applicable between 10^3 and 10^6 cycles. Method B may be used, but if you have a combination of pressure and thermal cycles, the current rules may be unconservative, because they consider each cycle type independently. The rules will be changed for the 2019 Edition to apply a Miner's Rule approach to combining these cycles. I have evaluated many components using Method B and find it to be sufficiently robust so as to capture potential issues with harmful details.

To summarize:
- > 10^6 cycles: mandatory full-blown analysis
- < 10^6 cycles, can apply "successful experience" unless poor fatigue details exist, then mandatory evaluation
- < 10^6 cycles, can apply screening Methods A or B with the limitations detailed in the methods. If they pass, then no mandatory analysis is required (the component is exempt from further fatigue analysis)
- Can always perform an analysis to demonstrate acceptability

 

[Paulettea]

TGS4, There is a point that comes to mind by thinking about what jtseng123 has described here. I think he is right when he wants to know what is so especial about 10^6 cycles. In the S-N curves provided in Annex 3-F there is no endurance limit (at least I cannot see any). I mean even if there is 10^8 cycles you can find the corresponding stress to failure. In many S-N curves for ferrous materials some endurance limits can be found at 10^6 or 10^7 cycles or somewhere between. But here 10^6 cycles means nothing.

Another question is what can be described as a cycle or a significant cycle when you want to check to see if there is more than 10^6 cycles or not. There are many pressure fluctuations in some equipment (e.g. after reciprocal compressors) that cannot be avoided. However, the range of this fluctuation in comparison with the maximum operating pressure is some times more important than the number of these fluctuation cycles. It is very well considered in methods A or B of the screening criteria. Unfortunately there is nothing mentioned in about how to calculate the number of cycles in screening method for comparison with a successful experience.

I mean as jtseng123 has pointed out there may be more than 10^6 cycles in the equipment but are they all significant cycles? Is it necessary to consider some criterion for significant cycles in 5.5.2.2 the same as 5.5.2.3 and 5.5.2.4?

 

[TGS4]

The fatigue screening/exemption methods are generally valid up to 10^6 cycles. Beyond that they are not conservative. That's not to say that a component wouldn't pass a fatigue analysis, only that the screening methods are not valid.

Regarding significant cycles in the similar experience, the key here is that the service is similar. Any aspects of significance for the cycles would be adequately covered, on my opinion, by a health dose of engineering judgement.

But I take your point regarding the 10^6 limit.

 

[jtseng123]

TGS4,

Does Germany code, British code, or European code have the same threshold of > 10^6 that fatigue analysis must be performed disregard experience for the similar items ? (the vessel we are buying has 3 million cycles, only pressure fluctuation)

When the code (Div 1) only asks loads shall be included (UG-22(e) and U-2(g))without telling what approach shall be taken, end user has the choice to use a rational method. For example, stress analysis for skirt base ring and skirt opening from two major softwares, PV Elite and COMPRESS, will never get the same results since both softwares use different formulas. PV Elite use British code for skirt opening analysis, while COMPRESS simply applies area replacement method which I do not agree. For the skirt base ring, one program will say overstressed while the other says it passes. There are many discrepancies between the two for non-code analysis and people around the world are still using. (The funny thing is, if one fails, we will ask vendor to use the other program so it can pass.)

We also have fatigue analysis done by major FCC cyclone supplier who uses British code to supplement Div. 2 Part 5.

Without technical basis of 5.5.2.1(c), and if codes from other countries do not have this requirement, we will decide not to adopt it for our Div. 1 vessel.

 

[TGS4]

IF your jurisdiction accepts that, then that is certainly your prerogative under the current rules. Future rules may be different.

 

[BJI]

jtseng123, yes the other codes have similar limits for fatigue screening. Fatigue is not covered under division 1, therefore you have to address it by other means (not discount it). This brings you to applying the requirements of division 2 for fatigue, and since your number of operating cycles exceeds 10^6 you are required to perform detailed fatigue analysis. I agree with TGS4, your AI is wrong.

However, I disagree with TGS4 regarding required evaluation. In this context, it is 'when evaluating experience with comparable equipment operating under similar conditions...' therefore the criteria for comparison as being suitable experience must include the same possible harmful effects. Evaluation does not specifically require calculation, even in the code context. TGS4 can you provide anything to support this interpretation, other than the opinion of one committee member? Obviously the number of cycles would still have to be less than 10^6.

 

[TGS4]

The words in the Code are as they are. If anyone wants a formal interpretation, then please submit a formal Request For Interpretation from the Code Committee.

As far as "analysed", "evaluated", and "considered" terminology goes, unfortunately, the Code does a poor job of explaining itself in that context. The Code writers have specific intents, but that intent appears to have not been communicated adequately - definitely room for improvement.