PDA

View Full Version : Metal Fatigue in Tube Frames?



Eric Marsh
11-15-2011, 04:44 PM
I am reading about best practices for maintaining old aircraft and it got me to wondering if fatigue is an issue in tube frame construction aircraft. I would think that they would be tougher than aluminum construction if for no other reason than that steel will flex more before failure but that's just a guess. Does anyone have any input on the subject?

Mike Switzer
11-15-2011, 05:21 PM
Steel doesn't respond to loads the same as aluminum - aluminum will stretch a little bit with repeated loading, steel will not stretch unless the applied load is over the yield strength of the alloy.

(Yea, I know that is a bit oversimplified, but it has been a while since I took materials engineering & I don't feel like remembering a whole lot of theory)

Kyle Boatright
11-15-2011, 10:00 PM
Following up on Mike's post, and keeping it very basic:

Aluminum has a finite fatigue life, and aluminum structures suffer fatigue even under their own weight. Eventually, they will fail.

Steel (at least in theory) doesn't fatigue unless it is overloaded. The main concerns for old steel frames are corrosion and imperfect welds which eventually crack.

Matt Gonitzke
11-16-2011, 04:17 PM
Mike-

Neither steel nor aluminum is going to 'stretch' permanently (yield) unless the stresses are beyond the linear elastic range. Both steel and aluminum exhibit linear elastic behavior below the yield strength and plastic behavior above it, so I don't understand how what you've said is true.


Following up on Mike's post, and keeping it very basic:

Aluminum has a finite fatigue life, and aluminum structures suffer fatigue even under their own weight. Eventually, they will fail.

Steel (at least in theory) doesn't fatigue unless it is overloaded. The main concerns for old steel frames are corrosion and imperfect welds which eventually crack.

A structure with a static load on it (i.e. an airplane sitting in a hangar) is not going to suddenly suffer a fatigue failure and fall apart. Fatigue failures are related to cyclic loads, not static loads. I've attached a couple of SN curves (stress vs. number of cycles) from MMPDS-01, one for 4130 steel, and another for 2024 aluminum. Both materials will fatigue if the maximum stress is greater than runout stress notated on the curves. There is a relationship between the maximum stress, stress ratio (relating maximum and minimum stresses experienced during a cycle) and cycles. This is probably a bit simplistic, but hopefully you get the idea. The steel frames you speak of do suffer from fatigue, as the corrosion and weld imperfections create stress risers in the structure, which initiate fatigue cracks.

Someone please correct me if I'm wrong, but what Mike and Kyle posted conflicts with what I was taught about fatigue and materials in engineering school.
1099
1098

Mike Switzer
11-16-2011, 04:43 PM
What I stated was what I remember being taught in my materials engineering course regarding any kind of repetitive load on steel vs aluminum. I don't have time to dig out my class notes now, but it is the reason drag race engines use aluminum connecting rods while endurance racers use forged steel rods. After only a few runs you can actually measure the elongation of the aluminum rods.

While you can design a structure with aluminum that will never show any evidence of stretching in it's lifetime, you have to oversize the structure so much that you lose any significant weight advantage over steel.

If I get time I will try to dig out the notes on the subject.

Bill Berson
11-16-2011, 04:56 PM
As Matt shows, all metals will suffer from fatigue depending on the alloy, heat treat, surface defects, stress and more.
In general, a proper welded steel tube airframe will usually last for decades (60 years+ is not uncommon).
Aluminum tube frames, on the other hand, might not last as long because they are normally bolted or riveted and the holes are stress risers.
Bill

Kyle Boatright
11-16-2011, 06:51 PM
Someone please correct me if I'm wrong, but what Mike and Kyle posted conflicts with what I was taught about fatigue and materials in engineering school.



I didn't keep my notes from materials science. That class was over 25 years ago. ;-)

But look at what the graphs you provided show:

The fatigue curves for steel appear to be headed towards a limiting value of about 40ksi. So under a 40ksi load, steel will never fail, regardless of the number of cycles.

The fatigue curves for aluminum are still pretty steep at 10KSI. If you had the entire curve, it would limit out at zero, meaning that aluminum will eventually fail after only one fatigue cycle, with that cycle coming from the structure's own weight.

Mike Switzer
11-16-2011, 07:31 PM
I haven't found my notes (like Kyle I took it about 25 years ago, they are in one of about a dozen boxes my mother sent out when she decided to "clean" my old room and what was in it's proper place on my shelf is now in one of these &(%$^**& boxes) and I haven't done any structural work using anything but steel in 15 years or so, but a quick review of 2 of my texts refreshed my memory somewhat.

When designing with steel, you design to the yield strength of that particular alloy (with an appropriate factor of safety). In most cases steel will not deform permanently until yield strength is surpassed.

When designing using aluminum, because it will deform to some extent below yield strength, (and this is the same for most non-ferrous metals) you design to the published fatigue strength of the alloy, which is usually significantly lower than the published yield strength. Fatigue strength is usually based on a certain (very high) number of stress reversals to failure. Since the yield point of non-ferrous metals can be hard to define, yield strength is defined by the amount of permanent set, usually 0.2 to 0.5 percent of the original gauge length (testing a standard sample of the alloy in a test fixture). If you design an aluminum structure using a design limit in the range between the fatigue strength & yield strength you will, eventually have a failure, it is just a question of how many cycles it will take. Now computer software can to a certain extent predict when this will happen (so you can design it to happen after the warranty runs out :D and hope the calculations were right). We were taught to design to the fatigue strength unless you were willing to test to failure to determine the life of the part.

Matt Gonitzke
11-16-2011, 07:52 PM
Mike-

A drag race engine's life is measured not in hours or even minutes, but seconds. I think what you are describing is creep, which is related but a different issue. I suspect there may be other factors than fatigue life that are driving the design of the connecting rods in the drag car engine since the lifespan is very short. They could be sized larger for lower stresses, but that might result in them either being too big or too heavy. The steel rods are likely used in the endurance racer engines because they have to last longer. I feel like that's an apples-to-oranges comparison since race car engines don't have a lot in common with aircraft structure, other than perhaps both being made out of some alloy of aluminum.

Kyle-

"The fatigue curves for steel appear to be headed towards a limiting value of about 40ksi. So under a 40ksi load, steel will never fail, regardless of the number of cycles." True, but if the structure had a stress higher than 40ksi, this would no longer be true. Both curves I posted have the same trend, but obviously different numbers because they are different materials.

"The fatigue curves for aluminum are still pretty steep at 10KSI. If you had the entire curve, it would limit out at zero, meaning that aluminum will eventually fail after only one fatigue cycle, with that cycle coming from the structure's own weight." You are misinterpreting the chart. The bottom edge of the vertical axis is 10ksi...there is no data down there, as the lowest run-out is at about 17ksi. One cycle is off the left end of the chart; notice it starts at 10^3 cycles. You have the entire curve...it does not 'limit out at zero', the numbers 'run out' below a certain stress level, meaning that the test specimen did not fail before the conclusion of the test. You keep making it sound like the aluminum structure is going to fall apart due to a static gravity load, regardless of what it is or how it is designed; if it is collapsing under its own weight, it is insufficiently sized, with the yield stress or ultimate stress being exceeded at some location. That would be a static strength failure, not a fatigue failure. That is a design issue and has nothing to do with what material it is made out of. I cannot think of any situation in which the static ground load case is the critical load case for an aircraft.

You could design an aluminum structure with a maximum stress of less than 15-16ksi or so and it would also have a theoretically infinite life, or an equivalent steel structure with a maximum stress less than 40ksi like you stated, and you'd achieve the same thing. The fatigue life of either the steel or the aluminum structure is going to depend on the maximum stress in that structure in accordance with the curves I posted. You could design a structure for a given application made of either steel or aluminum for a certain number of cycles before failure, or for infinite life.

I think you should have kept your materials notes ;)

Matt Gonitzke
11-16-2011, 08:26 PM
Since the yield point of non-ferrous metals can be hard to define, yield strength is defined by the amount of permanent set, usually 0.2 to 0.5 percent of the original gauge length (testing a standard sample of the alloy in a test fixture).

Can you cite some sort of publicly-available source for this? That is not in any of my materials or structures books, or notes from my materials class or 10+ structures classes in two aerospace engineering degrees. What you are saying doesn't make sense with how allowables are defined. For instance, if an allowable is A-basis, then at least 99% of the population exceeds the statistically calculated property value with a confindence of 95%. (source: MMPDS) If a material is always yielding below its yield stress allowable, then the allowable is bogus. A-and B-basis metallic materials allowables are all defined like this, as far as I know.

Mike Switzer
11-16-2011, 08:30 PM
Matt

There is no question that aluminum WILL deform at values below the published yield strength, why else would fatigue strength be the published design standard for non-ferrous metals while yield strength is used for steel? Everything in my last post was paraphrased from 2 of my engineering texts, I don't think materials science has changed in ~25 years.

Mike Switzer
11-16-2011, 08:32 PM
Can you cite some sort of publicly-available source for this?

Mechanical Engineering Design, Shigley & Mitchell 4th edition

Kyle Boatright
11-16-2011, 08:40 PM
Kyle-

"The fatigue curves for steel appear to be headed towards a limiting value of about 40ksi. So under a 40ksi load, steel will never fail, regardless of the number of cycles." True, but if the structure had a stress higher than 40ksi, this would no longer be true. Both curves I posted have the same trend, but obviously different numbers because they are different materials.

"The fatigue curves for aluminum are still pretty steep at 10KSI. If you had the entire curve, it would limit out at zero, meaning that aluminum will eventually fail after only one fatigue cycle, with that cycle coming from the structure's own weight." You are misinterpreting the chart. The bottom edge of the vertical axis is 10ksi...there is no data down there, as the lowest run-out is at about 17ksi. One cycle is off the left end of the chart; notice it starts at 10^3 cycles. You have the entire curve...it does not 'limit out at zero', the numbers 'run out' below a certain stress level, meaning that the test specimen did not fail before the conclusion of the test.



The point is that you can design a steel structure which will never fail. As the 4130 graph shows, if you keep the stress under 40ksi on 4130, it will never fail, regardless of cycles. An extrapolation of the aluminum graph will show that an aluminum structure will eventually fail, regardless of the load applied. And no, the test samples didn't necessarily fail, but they didn't run an infinitely long test, either - it isn't practical. It is certainly possible to design an aluminum structure with a lengthy and predictable service life, and that is often the best compromise in aviation. But that well designed structure will eventually fail.

Mike Switzer
11-16-2011, 09:12 PM
If a material is always yielding below its yield stress allowable, then the allowable is bogus.

This is why published yield strength is not allowable (as a design value) for aluminum structures - fatigue strength is, as the cycles required for failure at that level will probably never be achieved.

Eric Marsh
11-16-2011, 09:17 PM
Thanks for all the great info. My understanding is that aluminum rods are used in some drag racing engines because they absorb some of the shock of detonation which on some motors, especially those running fuel, can make the difference between bearing life and death. My airplane has 53 years on it's tube frame so I guess that means I've got another seven years left. :-) I was really curious about this because at first I thought that aluminum construction was the only way to go but after buying an airplane with tube construction I'm changing my mind. However, since this is kind of new to me I didn't know if fatigue would be an issue. What I'm hearing is that unless there is a real rust problem then cracked welds are the biggest concern.

Matt Gonitzke
11-16-2011, 09:26 PM
Well, I stand corrected...I found endurance limit explanation for non-ferrous metals in Bruhn. Not sure why it isn't in any of my newer books, but it wouldn't be the first time I've found an older book more useful than a new one:eek: Oh well, at least I learned something tonight.

"This is why published yield strength is not allowable (as a design value) for aluminum structures - fatigue strength is, as the cycles required for failure at that level will probably never be achieved." I agree...no airplane is designed for one cycle. I think what I've been trying to say all along is that a steel structure may also have a limited, finite fatigue life if the stress level is high enough, just like an aluminum structure. I wish I'd have thought of that a few hours ago, as it would have saved us all an evening...

Mike Switzer
11-16-2011, 09:42 PM
Matt, I'm curious, I graduated from Rose Hulman in 88, when did you graduate? I'm wondering if they are teaching fatigue life somewhat differently now - I haven't been involved in the automotive field for maybe 15 years, but even then I was seeing some trends in design that I & some of the other older engineers didn't agree with, designing components to values significantly over fatigue strength for both cost & weight savings, and the beancounters & six-sigma idiots didn't care as long as it made it past warranty & was not what the lawyers would consider a "life threatening" failure mode.

Like I said, I'm wondering if this area of design is being taught differently now? I never designed anything for consumer use that operated in a range over fatigue strength, but there were some things for military applications & for that big 500 mile race in Indiana that were on the ragged edge of yield strength & we hoped they made it....

Mike Switzer
11-17-2011, 07:44 AM
I was really curious about this because at first I thought that aluminum construction was the only way to go but after buying an airplane with tube construction I'm changing my mind. However, since this is kind of new to me I didn't know if fatigue would be an issue. What I'm hearing is that unless there is a real rust problem then cracked welds are the biggest concern.

Eric - the design I am working on will be welded steel tube - I chose that because long term maintenance will be easier & I am more comfortable doing the calculations for that type of structure.

Matt Gonitzke
11-17-2011, 03:47 PM
Mike-

I graduated this spring. Is your degree mechanical engineering or aerospace? I suspect an ME degree would go a bit more in depth into fatigue, as we had to have as many or more aerodynamics classes to be more well-rounded. Essentially, we had problems where we would determine the fatigue life of some notched-specimen shaped thing, given its material properties, geometry, and loads. Rather academic and not much of a real-life scenario, for sure.

Most of the focus was on static strength analysis; positive margin at ultimate load for no failure, and no permanent deformation at yield for static loads. Obviously, the sizing of many aluminum parts of an aircraft structure will be driven by fatigue life and not static strength, but we used the static strength and those allowables for initial sizing. There isn't enough time in a semester-long design class to do both the detail design and analysis, and the analysis probably would have been of no interest to those people in the class that didn't want to become stress engineers. Hopefully by the time I need to do the stress analysis on the aircraft I am designing I will have had enough training/exposure to fatigue analysis methods through my day job to comfortably do so.

Mike Switzer
11-17-2011, 06:24 PM
Matt - I am a ME. I will openly admit my math background wasn't good enough for the higher level aero classes. :D
A couple of my friends were ME/Aero, and both the required 300/400 level classes & the electives were different. I'm going off memory here, but I believe we were on the same track thru Statics & the first Dynamics class (Sophmore year) but we went separate ways from junior year on. The basic aero course I took was a 300 level course, as were the ME track courses I took that got into advanced design & fatigue, etc.

We had quite a bit of real world stuff in class, as our professors had quite a bit of experience - the Aero prof had been a Boeing test engineer, the materials guy was one of the designers of the Pershing missile, the engine prof had been an engineer on the Dodge Ramcharger racing team, and at the time at least a third of the ME department had worked at some point at the Langley research center.

Morrie Caudill
11-17-2011, 10:40 PM
I am reading about best practices for maintaining old aircraft and it got me to wondering if fatigue is an issue in tube frame construction aircraft. I would think that they would be tougher than aluminum construction if for no other reason than that steel will flex more before failure but that's just a guess. Does anyone have any input on the subject?

All the respondents below are correct to different degrees. You are concerned about a welded, steel, tubular, construction. Fatigue cracks initiate at points where high stress is concentrated and usually that occurs in the welds. To begin, the weld itself has the structure of a casting because it solidified from a molten state. Typically cast structures (welds) are about half as strong as the wrought material they hold together. Most fabrications are designed based on the strength of the welds if they will experience bending or tension loads. The shape of the weld is important. The point where the weld blends into the parent metal, at the edges and where the weld starts and stops, must be smooth. If the "toe of the weld" looks like it laps over and does not blend in smoothly, this is likely an indentation that will be a stress concentrator. Fatigue cracks like to start in these points. Close inspection of the welds for corrosion and stress points is all that's necessary. And if you find any suspect spots, blend them in with a file or grinder.
I haven't run into any welded, aluminum tubing structures in the airplanes I have worked on. They have all been riveted assemblies where the rivets are loaded in shear. Both structures, steel and aluminum, are usually designed to be stiff enough to resist bending, and fatigue cracks are not a problem.

dsbrantjr
11-18-2011, 10:31 AM
A bit off-topic, but Jimmy Stewart and Marlene Dietrich starred in a movie based on this subject, " No Highway in the Sky": An aeronautical engineer predicts that a new model of plane will fail catastrophically and in a novel manner (the tail falls off!) after a specific number of flying hours. There are some interesting scenes of aircraft fatigue testing.

Aaron Novak
11-18-2011, 12:45 PM
All the respondents below are correct to different degrees. You are concerned about a welded, steel, tubular, construction. Fatigue cracks initiate at points where high stress is concentrated and usually that occurs in the welds. To begin, the weld itself has the structure of a casting because it solidified from a molten state. Typically cast structures (welds) are about half as strong as the wrought material they hold together. Most fabrications are designed based on the strength of the welds if they will experience bending or tension loads. The shape of the weld is important. The point where the weld blends into the parent metal, at the edges and where the weld starts and stops, must be smooth. If the "toe of the weld" looks like it laps over and does not blend in smoothly, this is likely an indentation that will be a stress concentrator. Fatigue cracks like to start in these points. Close inspection of the welds for corrosion and stress points is all that's necessary. And if you find any suspect spots, blend them in with a file or grinder.
I haven't run into any welded, aluminum tubing structures in the airplanes I have worked on. They have all been riveted assemblies where the rivets are loaded in shear. Both structures, steel and aluminum, are usually designed to be stiff enough to resist bending, and fatigue cracks are not a problem.

I aggree with most of this, except the touching of a weld with any kind of grinder etc. Welds are NEVER to be ground, dressed, covered up etc unless it is for bolt head clearance etc. There are multiple reasons for this, and if needed they can be elaborated on later.

Bill Berson
11-18-2011, 01:11 PM
Fatigue cracks in small lightly stressed personal planes are mostly limited to vibration around the engine such as the engine mount and exhaust.
I can't recall weld cracks on the aircraft itself, in decades of inspecting. Some cracks from hard landings around the gear.

In general, any reasonable weld will be good on a homebuilt. Some of the standards that apply to nuclear plants and spacecraft are not applicable to homebuilding. AC 43.13 is a source of info such as grinding, mentioned above.
The EAA Museum has some samples of good and bad welds on display in a case. The bad welds look really bad.
A reasonably good looking weld will be fine even if not perfect.

A good weld is always better than a bad and undetectable composite or wood bond joint, in my opinion. Don't fear welds.
Bill

Bob H
12-03-2011, 10:01 AM
The original question was about whether a steel tubular frame would be prone to fatigue failure and the answer is "not likely" because of the high endurance limit of steel compared to aluminum.

A tubular structure is designed for stiffness which is a function of modulus of the tubes, steel at 30MSI and aluminum at 10 MSI. The stress level at which the structure is designed to operate is kept far below the yield point or proportional limit. But if you keep design stresses so low that fatigue never can happen, the plane will be very heavy or not fly. Most commercial transports have a warranted life where no primary structure will suffer fatigue cracking issues, somewhere around 60,000 hrs or more. To verify this, periodic inspections are done on critical areas using ultrasonics, dye pen or X-rays.
On a GA plane, the design stress levels are much lower so fatigue of aluminum structure is not an issue unless there are localized high stress areas or poor workmanship which allowed stress raisers to initiate cracking. This is why edges of detail parts must be smooth and blended and holes deburred.

If you have a steel frame and are concerned about fatigue, clean the frame down to bare metal, examine the tubes with a 10X loupe for cracking, check closely at welded joints for crack initiation, and if all looks good, prime and paint the frame to inhibit corrosion/rust.
Bob H