Material strength and fatigue

Recent postings have suggested that material strength and fatigue issues are poorly understood.

Fatigue is probably the single most important design consideration in any structure that experience cyclic stresses. It is also the single most significant factor in the long term reliability of a vessel. Yield strength is secondary to fatigue strength in any dynamically loaded structure.

Historically some scantling rules have been woefully inadequate when applied to modern composites because of the presumption of fatigue response based on minimal testing.

It is poor understanding and application of fatigue cycle limits that lead for example to keels falling off, chainplates snapping and major rigging failures even when large factors of safety appear to have been applied. The modern trend to lightweight boats is a mine-field of fatigue based failure as these vessels accumulate stress cycles with age.

The fatigue strength ( or SN failure curve ) is a plot of the number of cycles at a corresponding stress that would lead to failure. SN curves from material suppliers must be regarded with great care, especially in the case of composite materials where differences in fibre type, matrix- reinforcement bond strength or less than ideal construction methods (real world) and age can introduce significantly poorer results than the ideal samples tested.

Recently we have seen more detailed and accurate fatigue cycle testing of materials previously poorly studied such as wood-epoxy laminates as they are investigated for the production of wind turbine blades. These studies have been more rigorous than past studies since wind turbine blades are rotating and the number of cycles over the lifespan is high and the blades are large and expensive

Many people learn to do stress calculations and can convincingly calculate a required sections for a given load but completely ignore fatigue strength, or they hope that a factor of safety will cover fatigue. Sometimes this works, at other times it introduces a fatal design flaw because the material may not be as resistant to fatigue as the designer would like to believe.

For a reliable design we estimate the number and magnitude of fatigue cycles with an added safety factor, the component is then designed so that the experienced stresses fall below the critical stress for the number of expected cycles.

For example ( from Fig 1) a fibreglass component expected to endure one million stress cycles should be designed so that it experiences only 20% of its yield strength. That is the material should be 5 times thicker than the material strength would indicate for it to fail at around one millionth cycles. Now you need to add a safety factor. ( Old well used fibreglass vessels start to look a little suspect unless heavily built ).

Many racing boats chasing the lightweight rewards of speed are built with a safety factor close to 1, because strength and weight are on opposite sides of the coin and speed is easier and cheaper to achieve with low weight. The huge multi-hulls that have failed so spectacularly have been fatiguing to failure in a very short time under ocean induced stresses.

Designers particularly need to be aware of fatigue issues, be aware that there is a very well specified design life for some materials and should be able to use a SN curve along with a strength rating for a material.


Notes on Figure1

This is a generalisation of a very complicated subject, it illustrates a general relationship between material properties.

The trend line is taken from many samples run concurrently.

 

Very interesting post. May I ask how one might apply fatigue cycles to a non-moving object such as a hull?

Also, as usual with composites, there are too many variables to take the information too seriously. Composite longevity is highly dependent upon your choice of resins. Without knowing the resins and the ratios being tested, it's hard to know whether the data cited is an issue with composites, or if it lies solely with the resin being tested. It's known that certain resins degrade over time while others are more stable. It's also known that resin ratio can have a substantial effect on a part's final strength and its resistance to fatigue. Afterall, the glass stays pretty ductile for a long time in fiberglass (glass is a liquid, remember, not a solid)... but resins will store microscopic cracks in perpetuity... so reducing resin content and creating thinner layers of resin ought to improve fatigue strength.

Finally, here's an article that you "fatigue geeks" may found interesting. One out-of-the-box thinking aerospace engineer has decided to replace the hinges on his aircraft's control surfaces with nothing more than a sheet of aluminum, using its flexibility to provide deflection...

http://www.zenithair.com/kit-data/ht-aileron.html

 

WAVES

Mike, How does your numbers for wood/epoxy compare to Gougeon's?

 

So how do you figure waves? Is there an accepted "number of cycles (waves) per minute"?

 

To add to Toot's point about composites;in addition to choice of resins there are factors such as the working environment and the ability of the laminators.
In respect of the use of flexibility to provide movement,Formula One racing cars have been using flexures to connect the suspension to the chassis for some while.

 

Yes. I've started saying: Composites don't provide strength, they provide the potential for strength.

I don't think it's an exaggeration to say that 90% of a composite's strength comes, not from the materials, but from the fabricators themselves.

 

on this point there is a well known aircraft crash, boeing model, the rear pressure door failed because of a failed repair patch a specific test showed that the patch failed exactly at that many cycles due to only one row of rivits allowing the patch to flex. i forget the exact incedent. but the time tested method of finding out is building a chunks of hull and testing it to destruction if ya really want to find out......

 

Raggi

Don’t get too hung up on the numbers, this was a materials comparison. The point being you should design to a reliable SN curve.


The report is long and detailed but remains copyright material. Thje synopsis circulating also cites commercial copyright. The timber in the epoxy timber panels was African Mahogany chosen as a particularly strong fatigue resistant timber, various ply thicknesses numbers and orientations were tested for the composites. Each panel type was tested for tensile yield strength and the fatigue characteristics plotted as a % of that strength.
The resulting normalised curves are surprisingly similar , all ending up very close together for the higher fatigue cycle range.

cheers
---------------------------------------------------------------

Toot
I’m getting old and irascible so this might be a bit personal.

Why is it that only ignorant people patronisingly refer to learned people as Geeks?

In the future try and do a little research before posting, rather than shooting from the hip with half baked ideas. You are apparently a layman, you obviously know very little about either material science or design loads on hulls and perhaps have some vested interest in maintaining a prejudiced view re GRP properties? Too much yard knowledge is urban myth, and facts trickle slowly into the consciousness of society.

You are correct in your assumption of strength WRT lay-up quality but there is a limit to that strength and that is the ideal panel. Scientific testing gets as close to that ideal as possible.

With 20 year guarantees on 50 foot turbine blades 100’s of millions of dollars are at stake, the research is done properly. If you are going to the time and expense to fatigue test then you don’t randomly approach the exercise, materials science is more thorough than you seem to think

GRP is well enough known to optimise the fibre-matrix ratios for fatigue response. GRP fracture mechanics is well documented and well studied, the SN curves shown are optimal, the real performance will be considerably worse with matrix age and water absorbtion. Note that all the testes are on laboratory produced ideal specimens.
Labs will produce panels in far better control than any yard could hope to.

These tests concur with other accurate controlled tests carried out by other institutions. Understand that they are undertaken by people who know more about the subject than you can guess at (as your postscript implies).

Secondly
Room temp glass flow is urban mythology long since put to rest. Glass behaves in all characteristics as a brittle solid, and to classify as anything else only misleads. The myth unfortunately endures amongst laymen. In ambient temperatures it is not ductile or viscous it doesn’t self relieve stress and it cracks unremittingly and with ease from any stress raiser since the brittle material is unable to yield. For future reference for fracture mechanics it is an amorphous solid.

Thirdly
Hulls go through a myriad of stress cycles. Please read up on the subject.

Don’t post opinions as facts, precede the post with “I think” or IMO , ask questions such as “Isn’t it shown that……….” Incorrect statements of fact don’t help anyone.

 

I did some research because somebody wanted a professional race boat that lasts only 3 seasons. I found the following webpages:

http://users.ugent.be/~wvpaepeg/fatigue/composites.html

Basically this man sais that SN curves don't really work for composites. The reason is that the damage in composites that occurs due to fatigue reduces the material stiffness. Older boats are less stiff. As a result highly loaded areas attract less stresses over time and loads get re-distributed.

The reason fatigue is not that important in boats at the moment is because of relatively low cycles. Looking at the chainplates and assuming a fatigue life of 20000 cylces (very low) and 1 tack every 10 miles you would get a life of 200000 miles! It is different in wind turbines which might clock up 1000000 cycles in 2 weeks.

 

Toot, Gougeon brothers write that stress from waves has been measured to occure every third second. They also estimate that 833 hours or one million cycles (10^6) means four years of seasoning week end sailing.
The Gougeons have a figure that looks a bit like the one above, but they use the term "percent of ultimate strength" or "fatigue strength as percentage of static strength". At one million cycles they claim that wood/epoxy is at 60%m, the same as carbon fibre composite, while steel and aluminium is approx 40% and s-glass/epoxy is 20%.

 

Look on the Ship Structure Committee website for reports on how fatigue is handled for ships and marine structures (www.shipstructures.org - and a number of good reports on composites in general). This sort of analysis is pretty much routine for military and commercial ships, (just did one this week) or built into allowables in codes like NVIC 11-80. However the basis strength for metal S-N curves is ultimate tensile.

Composites are an interesting issue, and there is some additional material on composite fatigue also available on the web, mainly at Colorado State U (I think - google it).

 

Nice thread-

How is it thought that this area of research/analysis applies to the older fleet of FRP sailboat hulls in use? Many of these vessels are entering their 3rd and 4th decade of service. Seems to put a new light on the myriad broker adds which note "just back from circumnavigation", "just crossed the Atlantic and ready to go again". Should these histories really be read as notations which simply state the degree to which these vessels finite service life has been used up? I work in the marine trades and really don't see warning signs of hull and deck problems suggestive of generalized fatigue failure. Many many damaged wood core decks, some tab work needed, rotten bulkheads and the like, but not really the former case. I do see evidence in the power boat sector: catastrophic hull failures and the like. I would venture that these vessels see a higher load and exhibit failures as a result, even though the cycles are lower.
My interest is really in the cruising sailboat fleet. Is there a load amplitude which is so low as to allow a infinite number of cycles from a practical standpoint. Or to state this question in a different way: are hull stresses understood well enough to place load numbers on a given hull panel as expressed in a partucular hull form? Numbers which could then be used to determine service life at a given load amplitude?

 

MikeJohns-

As for "geek", you are absolutely right. I misspoke. What I meant to say was "nerd". Either way, I didn't mean it in a patronizing way. Besides, you can't ignore the fact that I, myself, am here and responding to the thread so, at worst it's a pot-kettle sort of thing. My circumstances have been such that I grew up with computer nerds/geeks, most of whom went into engineering and computers and other intellectual fields and did quite well for themselves. It wasn't meant to suggest any social awkwardness. But, on the other hand, here we all are on the internet when we could be out shtooping our wives or girlfriends (or boyfriends, not that there's anything wrong with that....). So maybe the shoe does fit afterall.

As for glass being liquid, a little additional research on my part indicates that you are correct. It seems the proper term is "viscous solid", rather than a liquid. Learn something new everyday, eh? In my defense though, I learned that tidbit from a highly-respected Civil Engineer with over 40 years of practical experience. I guess that's what you get for listening to "experts".

And I'll tell you. I looked ALL OVER your post, for some reference to suggest the level of quality used to generate your chart. I looked for a laboratory's name. I looked for a large company to have funded the research. I looked to see if, perhaps, it was created by a company that makes metal blades, or steel parts (bias?). I checked for margins of error, perhaps a scatter-plot of the test samples- I wondered whether the GRP samples had a greater range of values (indicating fabrication skill/accuracy may have been a factor). I checked for a footnote regarding the resin-ratios and fiber orientations for the reinforcement used. I checked to see whether the modulus of the resin used was listed. Epoxy? Polyester? Vinylester? Prepreg? Wet Layup? Infusion? I found none of this! So you'll have to forgive me for initially viewing your chart with substantial skepticism. It certainly isn't ready for "prime time". That isn't to say it's not useful for the purpose it was initially intended. I'm just saying that I have questions (not doubts, mind you, but questions) as to its applicability.

Nevertheless, I certainly welcome all this sort of information and any additional information you can give which may help make the chart even more relevent and convince me that it is applicable and accurate to composite manufacturing as a whole would be greatly appreciated by myself and, I'm sure, many others. In the mean time, it's just a pretty picture with something which certainly ought to bear more than merely notional value to the casual reader. But surely, you can appreciate the risks in taking any information from a source, no matter how reliable, if it isn't backed up by the necessary documentation, calculations, references, or footnotes, etc... FWIW, I've seen many charts like the one you presented. By and large, the information rings true. But that doesn't prevent me from having many many questions about the methods employed in its creation.

I don't doubt that hulls go through stress cycles. I merely was asking whether there was some sort of "standard assumption" for such. "1 for every 3 seconds" was all I was really looking for.

Anyway, I saw that later in a follow-up post, you provided some of the information I was seeking and answered some of the questions I had. I appreciate that.

As for "shooting from the hip". I was trained and worked as, among other things, a composite tech in aerospace manufacturing. So you'll have to forgive my admittedly-weak knowledge of boat matters per se. But, honestly, I do try to look things up before I ask silly questions. I didn't think the asking about cycles on a hull was silly in the least and, in fact, was having troubles digging up the information which, I felt, ought to easily be on the top of some member's head.


Cheers!

Ben

 

I think many older grp sailboats are designed and built so strong that the stress in normal conditionas (waves, hull motion, rigg loads) is allways far lower than, say, 20% of the ultimate strength. Fore example, I remember an old (1977?) Swan 65 that made a big hole in the side of a car ferry (9mm steel?).

Then you have racing boats like the Soling, they get soft after one or two years and are sold for a few thousand euros/dollars.

 

Karsten
fatigue is very significant, hulls can be very highly stressed and at sea very quickly accumulate cycles on the same tack from wave induced accelerations in roll pitch and yaw and pressure variation.

I wonder where the fatigue failures fit into this simplified view? I have seen major structural failure of GRP vessels through fatigue, often relatively new vessels. Problem being we have no rules for non commercial vessels and factors like fatigue get left out or blithely ignored by some yards because they read a comment like that which you posted. As I said before Urban mythology abounds even amongst skilled workers. That is why I posted this material.

I would like to see some lifespan predictions for many of the modern expensive light weight cruising boats!


Ben (toot)
ok

We have to post simple generalised information since this is a boat design forum andmeterials science lectures are not very exciting.


Thats why I wanted to stick to general information, otherwise the educational value is lost, in marketing all you have is one glance to arouse enough interst to absorb the message. The complex detailed curves of which I could post hundreds would confuse and obfuscate the issue.

Here's an example


Now add ageing water absorbtion and a less than perfect layup....