Residual fatigue life assessment of structure with unknown loading history 1

[rather_be_riding]

Someone floated a project the other day which involved refurbishing a used item of plant for a customer who has no loading history for it. The plant structure can be considered to be fatigue sensitive and some anecdotal evidence suggests that maybe at some point someone was actually concerned about this but there's really nothing to work with.

For the purposes of debate, lets say that the structure all checks out and welds are subjected to NDT and turn out to be in good condition. How do you go about determining the remaining fatigue life? It could be 90% of its way through its design life or 10%; at either point, the structure should still be in good condition. I recognise this is probably a somewhat unusual case and maybe no one has a great answer but even a good text reference would be appreciated.

Side note: I was unsure if this was better here or in structural but figured that maybe the aero engineers might have valuable input and they'd be more likely to monitor this forum.

 

[Jboggs]

If all the tests indicate that the item is in fairly good shape it could be offered to the customer at a discount price with the provision that there is no warranty of continued performance due to the fact that past history is unknown. If the customer balks that that, quote them the price for a new item with guaranteed zero usage. Give them all the information required to make an informed decision and assume as much risk as they are willing to take. Let them make the choice. If you share everything with them they could not accuse you of hiding any critical information.

 

[desertfox]

Hi

If there is absolutely no data and there is no idea of what the loadings are, I don’t know where you would start. That said an old rule of thumb which you may or not be aware of:- Stresses in a structure do not usually cause a fatigue failure provided the working stresses do not exceed 25% UTS of the material. So if by chance, some loads that the plant would be subjected too, or could be reasonably estimated then one could run a check for stress level and check it against the UTS. The downside is the design usually ends up much larger than it really needs to be.

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

 

[rb1957]

maybe cut a piece of flange, maybe with a hole in it, maybe "virgin" and drill a hole, and run a fatigue test on it and compare the result to new material (ie book values and/or run a test on new material).

another day in paradise, or is paradise one day closer ?

 

[rather_be_riding]

Thanks so far.

@Jboggs - not an option. It's registered, has to have a lifespan etc. It's one of the few cases where our workplace safety government bodies do get involved and is a definite killer if someone screws up the life assessment. Once it's classified, it SHOULD be cross-checked on a yearly basis to make sure the utilisation matches the original classification etc. I can understand the end customer's desire. There's a million dollar piece of gear standing there, looking for all the world to be in good shape, getting wasted. And for all I know, it's got years of life in it. I just don't know and I'm not sure how anyone can find out.

@desertfox - I'm aware of the endurance limit. Without undertaking a detailed review of the design, I'm 99.8% sure a fairly significant number of components will be stressed well beyond it. I'm extra confident of this because at some point someone significantly 'downgraded' it, which is almost always done in an attempt to squeeze under that limit.

@rb1957 - that's a great idea and I hadn't thought of it. It may not be practical in this case because a) the bits that are likely to be heavily worked might be rather hard to cut out and b) without good utilisation data, it can be rather difficult to determine what's been worked hard (for example it is possible to heavily work one area or component whilst leaving the remainder in almost 'as new' condition). However, it merits much more thought. Maybe I need to take a bucket load of samples which would correspond to a lot of repairs to the sampled locations thereafter. Maybe I can find a good canary component which is fatigue sensitive and always loaded. Regardless, it's a potential approach which could yield enough data to make a proper scientific call on the issue.

 

[MintJulep]

Might be worth getting a copy of this paper.

https://journals.sagepub.com/doi/abs/10.1177/1056789519860242?journalCode=ijda

 

[rb1957]

yes, I understand no historical spectrum of loads. But if you test a piece and it fails at 50% of the predicted number of cycles, well, you know something ... just with not much confidence !

If this new application is fatigue critical or fatigue sensitive then your situation is very problematic.

Another approach, from my "wheel-house", would be DTA ... develop an inspection program that'd ensure cracks would be found before the piece failed. You can avoid the unknown history by assuming a just-not-detectable crack is present; inspect now, then good for a "repeat" inspection interval. Not at all easy to do, but another approach.

another day in paradise, or is paradise one day closer ?

 

[rather_be_riding]

@MintJulep - believe it or not, even prior to your message, I was doing my level best to short of paying for it but so far I haven't found it included in any of the subscriptions I have access to. Strange. I've read a couple of other thermography/fatigue papers and abstracts and the feel I get is that it may not be practical for larger plant where our loads might be in the realm of 50 tonnes or more; the excitation might well need to be of similar scale to demonstrate detectable results and might be difficult to achieve in the real world. I might have to just cop it on the chin and pay for it; it just stings more when you do and it doesn't help . That said, I'm super interested in whether there's a ND means of assessing such things. If it was relatively reasonable in price, I'd consider investing in adding it to more common place inspection work to build up a personal database of cumulative damage results for similar equipment. Extra useful if I can correlate it to cases where I DO know the history.

@rb1959 - I might not have been clear enough; I wasn't referring in my reply to just the lack of load history but also the specific lack of per component load history. Normally what we assume is that each load cycle loads each component in a worst case kind of fashion. However, when we're working backwards like this, we must almost make the opposite assumption to be conservative. IE - we can't test a convenient sample only from component A and assume it to apply to component B in the case that movable loads (and the idiosyncracies of the previous usage) might have managed to only really load component B leaving A in perfect condition. That being the case, you end up having to test a lot of different areas. Not impossible, just needs to be allowed for. It's a good idea.

The application was and is definitely fatigue sensitive. Always needs to be considered although, quite often, the equipment may be so under utilised that it doesn't actually apply.

I must profess lack of familiarity with the DTA acronym but I get the general picture. I've employed that approach once previously on a monorail in a foundry that couldn't stop working for the period it would require for the very unique replacement to be sourced, fabricated and installed. It was expensive and I didn't sleep much. Definitely not easy to do, I'd definitely employ a fracture mechanics consultant to advise and on a substantial structure the cost of undertaking the NDT on a regular basis will probably mean that it's unreasonable on a longer term basis (ie good to limp it through for a period if there's considerable financial incentive to do so, not so suitable for a 10-25 year lifespan).

 

[drawoh]

rather_be_riding,

How catastrophic will it be if this thing fails?

--
JHG

 

[rb1957]

if you don't know the fatigue load spectrum, how do you do fatigue analysis ? You could say 100,000 limit load cycles, or you say you've designed for infinite life.

But to understand the current material's fatigue state given an unknown history, yu can tet a piece under standard loads and correlate the result with material standards. This is not very good, in that confidence should be very low (how do you know your sample is representative ?).

Like drawoh, what are the consequences of fatigue failure ? what is this in any case ? (machinery ? structure ??) what material ??

another day in paradise, or is paradise one day closer ?

 

[rather_be_riding]

@drawoh and @rb1957 - loss of life would not be an unreasonable outcome although productivity and damage to plant are substantially more likely. And of course I'd rather avoid the blow to the ego, the reputational damage, the loss of registration and potential jail time . Conservative is the only approach here but I do think there's at least room to consider if there's an alternative option to the super safe 'nah - can't be done, mate' advice. I'd classify this as lifting equipment (which is why I'm involved). I'm expecting the majority of what I'm concerned about to be AS 3678 Grade 250 plate with maybe a 300 or 350 thrown in here or there. There's a possibility that there might be a few pinned joints or the like out of 4140 or the like. Technically that all has to be sampled to be sure though unless someone somewhere miraculously finds some drawings. Welds are highly likely to be high quality sub-arc on longer runs and MIG in the detailed areas; hopefully of high quality too - they should have fabricated in the first place with fatigue in mind.

You have a reasonable point about representative behaviour. The only way past that is multiple samples, from multiple locations and using them all to come to an overall conclusion where the weakest link will define the remaining life. Of course, we're also looking to ensure we have sufficient safety factor so 100k full load cycles to sample failure will end up being something more like 33k design life cycles combined with a more stringent maintenance plan than usual.

Even if this project doesn't go ahead (or I bow out due to excessive risk), I think this is a good discussion/debate to have as such things come up semi regularly in my experience. It's quite common that lifting equipment is massively underutilised but there's just no data to support such an assessment. The easy out is to say it can't be done but it's wasteful and if a solid approach to control the risk can be developed, I think it's almost a little unprofessional.

 

[CWB1]

As I tell junior engineers regularly, let the standard process drive engineering and take it one step at a time. Attempting to solve technical problems without first understanding what they are, how many exist, or their importance in the grand scheme is pointless and causes rookie mistakes. Create a DFMEA to identify and rate all possible failure modes, then perform whatever engineering analysis necessary on the highest scores to identify weaknesses, standards and acceptance criteria that must be met, and any functional testing necessary. Follow that up with a manufacturing PFMEA. As you're not starting with certified material, the PFMEA will obviously identify material testing on critical components identified by the DFMEA. The material either meets your requirements or it doesnt for any given component, but a lifecycle analysis completed XX years ago is irrelevant.

Engineering at a high level is all the same. Follow the process, not only does it CYA legally but also makes the myriad of project planning simple while minimizing risk. There may be a lot of expensive work to do on this project or very little, but you won't know until you get through the planning.

 

[3DDave]

Is there some failure mode where fatigue causes a structure to crumble as dust at N+1 cycles or explosively ruptures without showing any external cracks? For fatigue intolerant structures the typical approach is to make a decent guess about the smallest non-detectable crack and the pace at which that crack might progress under the known loading to set the period of inspection.

You don't need the drawings - they are a hindrance. You need to accurately reverse engineer the as-built item and have an FEA done to see which areas are subject to the highest stress and supply an estimate of what the critical crack length will be. The toughest part is identifying the material properties in the heat affected zones, particularly the alloy, but I suspect it can be done.

Strip it, dye penetrant inspect the whole thing for external cracks, maybe Magnaflux it; then do X-rays for the places where the FEA suggests internal cracks might originate or just to find where voids might exist.

There doesn't seem like there's a lot of risk - there may be a lot of cost.

 

[rather_be_riding]

@3DDave - nothing so dramatic. Inspection is difficult/time consuming/expensive during service. Catastrophic failure of such things in my experience is usually due to poor maintenance. The rub is that poor maintenance is more the norm than the exception.

At a guess, it might take a day and a half or two days to crawl over the structure to NDT it each time. That's not including any of the time required to strip it and recoat. A somewhat similar item of plant I was involved with briefly a few years ago had crack growth periods in the realm of a fortnight. The inspection costs were phenomenal but at least it didn't fall down.

I think you're right. Entirely possible to reduce the risk substantially with that approach. It just might be that the client doesn't like the price, particularly that of ongoing inspections.

 

[LittleInch]

There's a million dollar piece of gear standing there, looking for all the world to be in good shape, getting wasted.

This is your problem - not the fatigue life issue (IMHO).

What there is, ia a pile of scrap metal. If, as you state, this needs to be licenced/registered for a set period to be of value to anyone then without the past data you can't realistically provide the data to make an assessment, other than maybe 6 months or 1 year.

The same thing crops up on boilers and pressure vessels when someone comes on and says I have this PV, but the nameplate is missing and I don't have any paperwork to go with it. It's a crying shame, but the same answer, what you have is an expensive paperweight. It's the fault of the previous owner not to do the testing or provide the data, or maybe they never had it but weren't as carful as you are.

Do you have ANY history of this thing, even when it was made, where it was installed, was it inspected in the past - anything? If not or very little then I think it's time to walk away.

I used to visit Baku occasionally years ago and on every journey down the coast you went past a huge jacket on its side, ready built and just abandoned by the Russians when they bugged out. It sat there for years whilst everyone tried to find a use for it and if it could support anything worthwhile without collapsing. No real data or design or material certs worth talking about. All sorts of samples were taken and the design reverse engineered and in the end I think they scrapped it or had to accept it could only hold up a fraction of what it should. Cut your losses early is what I'm hearing.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[desertfox]

I think I am with LittleInch on this one in that its time to walk away, I remember I job where we were putting new equipment on old foundations and the client wanted where possible to use the existing foundation bolts even though they were 40 plus years old, hence to say all the foundation bolts were renewed, no way were we risking using foundation bolts we knew nothing about. The more (rather be Riding) tells us the more I am inclined to walk away, seems to me that someone wants a cheap job without carrying the responsibility

“Do not worry about your problems with mathematics, I assure you mine are far greater.” Albert Einstein

 

[dvd]

It seems as though a risk assessment is in order. This piece of equipment is not really different from a new machine which fails due to a hidden flaw. Define the protocol and procedures that will be used to monitor and ensure safety, and get the owners to commit to the safeguards and sign off on it. If you can't afford the price of the consequences, then move on.

 

[rather_be_riding]

@LittleInch, desertfox, dvd - fair and reasonable I reckon. That's probably where it will end up. I actually suspect that if I give them an estimate on addressing it the 'right' way, they'll chase around getting three quotes (which is reasonable) and one of those three will be a service provider that doesn't use an consulting engineer and just doesn't care. Lifting/handling equipment is full of such cowboys unfortunately. In not so distant times, there was a crane supplier who used to buy old equipment out of factories, cut and shut them to new dimensions, paint them and install them as 'new'. I suppose the quotes didn't explicitly state that the crane was new! Every now and then, I get someone with one of these nightmares referred to me and it would be great to be able to help them out in a manner that doesn't saddle them with excessive maintenance costs thereafter. Nothing to do with this project at all but similar kind of problem and application.

Re: history. I know where it stands is where it was installed. Way too big to move without someone definitely knowing about it. Sometimes, that means you can have a really solid educated guess at the use something has seen but it appears in this case noone can/will. I don't have a date of manufacture but would be surprised if we can't track that down somehow. There should be an original registration with the state body. The hard part there is usually just getting their assistance. It will have been inspected and maintained; to what degree is anyone's guess and the documentation is in the wind. So sweet FA on the whole. Without the loading history, all the rest is pointless at this stage anyway.

 

[LittleInch]

Well you can only hope the regulator is wise to these cowboys...[pre][/pre]

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[FacEngrPE]

Refurbishing a crane is to a large degree an economic question. From your statements above, the costs of investigating this crane to the point that the risks of future service are acceptable (at least as far as your firms's involvement is concerned) are to high to be practical.

From a strictly engineering standpoint, if the crane is designed to FEM or something similar, you are more likely to find fatigue life limiting situations in the machinery, and in structural locations with stress concentrations. Developing an inspection plan is important, the plan will help to identify the risks and costs. If the inspection plan is too expensive to be worth your effort you have your answer.