Metal Fatigue - Steel vs. Aluminum

[rsterne]

I ran across an interesting graph today.... It shows the stress for Steel and Aluminum below which metal fatigue will not occur, plotted vs. the number of cycles....



At 1 million cycles and beyond, steel will not fatigue if the stress on it is less than 30,000 psi.... Aluminum continues to fatigue, however, although more slowly than before a million cycles, and can still be subject to fatigue at a stress as low as 14,000 psi at a billion cycles.... To me, this means that using a safety factor of greater than 2:1 to yield on steel means that metal fatigue should not be an issue, but you would need 3:1 on aluminum if enough cycles were involved....

This has nothing to do with using a 3:1 safety factor as a basic design criteria, I am not arguing against that practice, and agree with it because it gives an additional resistance to things like stress risers (threads, notches, square corners, etc.).... Such stress risers can increase the LOCAL stress well above the calculated stress, of course, and cause metal fatigue and cracking starting from those features....

Bob

[Scotchmo]

The endurance limit will depend on the type of steel. Something like music wire or chromoly will be much higher than 30,000psi.

The important point is that with steel (and Titanium), you can design for essentially infinite life. Not possible with Aluminum.

[rsterne]

I couldn't find it, but the number I remember for steel is 45% of the yield strength for the endurance limit.... so yes, higher grade steels would have a higher endurance limit.... that is why the 2:1 safety margin to yield is a good guide for steel, assuming a polished finish and no stress risers....

Bob

[Cal]

The various commercial aluminum alloys have differing characteristics.  The 6000 series is noted for long service life.  Not so for the 2000 and 7000 series materials.

Generalizations are just that,  Choose wisely

[Scotchmo]

I couldn't find it, but the number I remember for steel is 45% of the yield strength for the endurance limit.... so yes, higher grade steels would have a higher endurance limit.... that is why the 2:1 safety margin to yield is a good guide for steel, assuming a polished finish and no stress risers....

Bob

45% of yield? Probably about right for very hard steel. 50% of tensile strength (not yield strength) for most steels.

If you use 45% of yield and then a 2:1 safety factor, no stress risers - that should be safe enough.

It's difficult to design with no stress risers. So they should be factored in first and then apply the safety factor.

[drumie]

Where would carbon fiber be in a chart like this?  Just thinking regarding CF tanks.

[Scotchmo]

Where would carbon fiber be in a chart like this?  Just thinking regarding CF tanks.

It is not so simple. The failure modes of a composite material would be much more variable/complex than the homogeneous steel and aluminum alloys.

[Trim Tab]

...

[rsterne]

Interesting chart for the composities.... However, it doesn't address the most difficult part for them, and that is figuring out the ultimate strength in the first place.... There are so many variables, fibre content and orienatation being the key ones.... since they are not homogenious, and the fibre strength lies along the fibre axis, with the resin doing little....

Bob

[Airsnipe]

What about titanium? If memory serves, it has fantastic fatigue characteristics. 

[moorepower]

Most tests are performed from 0 psi up to xxx psi. Wouldn't the fatigue be less from say 2200 psi to 3200 psi compared to 1000 psi to 3200 psi.?

[rsterne]

I couldn't find a fatigue graph for 3AL-2.5V Titanium, but I found several for other alloys, but only to 10^7 cycles.... The shape of the curve is in between those shown above for steel and aluminum.... It is more curved, but never reaches a plateau like steel, although maybe it does at higher cycle numbers, the trend is going that way.... Some of the curves look much like that for aluminum, but even steeper, where the tensile strength is down to a third of its original value after only 10^7 cycles, while others are much shallower losing only about 40% of their original strength over that cycle life.... In many cases, it looks like the curve is steeper than for aluminum for small numbers of cycles, eg. 10^4 or 10^5, which is the area we are mostly concerned with.... Losses of 30-40% of tensile at 10^5 cycles are common, compared to aluminum at about 25% in the graph above....

I have two comments....

1. There is a huge variation in the tensile strength depending on the cross section being tested, and whether or not a notch is present, even the surface finish.... and also a large variation between alloys.... Therefore, I wouldn't put too much faith in any graph unless the test conditions match your conditions....not only for Titanium, but for steel and aluminum above....

2. From what I have seen, any comment that "Titanium doesn't suffer from fatigue", or "has fantastic fatigue characteristics" is totally incorrect, and those making that comment need to produce data to support it.... I personally would treat it like aluminum, in terms of how it fatigues, from the limited data I have found in the MIL-HDBK-5 Materials Handbook.... It may, in fact, be worse, but unless someone comes up for data for the commonly used 3AL-2.5V we should avoid speculation about that....

I can't answer the question about the difference between a cycle from 0-3000 psi (like a SCUBA tank used for diving) and the same tank used down to only 2000 psi and then topped up to 3000.... but it makes sense that the cycle life would be increased.... The point is, that if you are pushing anything to the point of having to worry about cycle life, you better either have the hard data on the life expectancy (and not exceed it).... or you should redesign it so that the safety factors are so high it is no longer an issue.... IMO....

Bob

[Airsnipe]

Very interesting info on the titanium. I have always been told and read that grade 5 and 9 Ti is very much like spring steel and in fact has been used for springs. I know allot of myths about titanium but had a apparently misunderstood this. I have made pocket clips and other springs things from grade 5, and I can tell you it does make a pretty good spring but I have never put anything through the numbers of cycles addressed here. But I suppose just because it's springy, doesn't mean it has good fatigue properties. But I am going to see if I can find some info on 3Al-2.5V as now I'm curious. 

[Airsnipe]

Well I found this...

Site: http://smt.sandvik.com/en/materials-center/material-datasheets/tube-and-pipe-seamless/sandvik-ti-grade-9/

Ti-3-2.5 tubing has been used in aerospace hydraulic lines for over 20 years due to its high strength to weight ratio and its excellent fatigue performance. A typical stress vs cycles to failure (SN) curve for CWSR Ti-3-2.5 tubing is shown in figure 8. This data is from planar flexure fatigue testing of a 90° bent tubing sample with 4000 psi (275 bar) static internal pressure. The tubing size used was 3/8" (9.53 mm) OD x 0.019" (0.5 mm) wall.

[rsterne]

Interesting graph, but note that it starts at 10^5 (ie 100,000) cycles.... It also appears to me that the stress at 10^5 cycles is already down to 80Ksi, whereas for 3AL-2.V the tensile strength is quoted as 155 Ksi for CWSR grade.... The graph plateaus at 72 Ksi (at 2,000,000 cycles), which is 46% of nominal, placing the graph almost identical to that for steel (above) on a percentage basis.... The way I read that graph, if you need a cycle life of 2,000,000 or more, you should use a UTS of 72Ksi as your design point....

Bob

[lloyd-ss]

I think there might be a misinterpretation of the 3.5Al-2.5V Ti graph.  If I understand this correctly, this is a tube bent at 90 degrees, internally loaded to 4000 psi, and then it is repeatedly stressed back and forth across the 90 degree bend such that an additional 77ksi Bending Stress is induced in the tube wall (the 3 test point dots at the upper left of the graph). Converting bending stress to tensile stress can be imprecise, but I think adding the 77ksi bending is somewhat similar to pulsing the internal pressure to about 11,000 psi for 100,000 cycles. A serious torture test!

The one area that immediately comes to mind where fatigue life is a real issue in airguns, is in springer springs.  Those springs are loaded way, way past the fatigue limit, and often right at the yield limit (setting a spring ??), and therefore, they break.

Generally, you need lots of vibration (airplanes, automobiles, springs) or rapid cycling (high speed automated machinery, automatic weapons)   for the stress cycles to get high enough to be  a serious concern; or bad design with severely overloaded parts. 

I think keeping true to our 3 to 1 safety factors should generally keep us safe for the low cycle rates seen in airguns.

[rsterne]

I agree, Lloyd.... My main concern is for the reduction in tensile strength with cycle numbers in the thousands (which may happen).... Am I reading the graph for steel and aluminum wrong when I say that if the imposed stress at 10^6 (a million) cycles is less than 30Ksi for steel or 24 ksi for aluminum then fatigue is not an issue?.... and for the Titanium that is 72 Ksi?....

I always understood that if the imposed stress on steel was less than 45% of the yield strength (roughly that 30Ksi value) then an infinite number of cycles was possible.... As the imposed stress increases towards the yield point, then fewer and fewer cycles were required before failure (cracking) occurred.... It would appear that Titanium has a similar curve, with a plateau at about 72 Ksi, so if we design well under that (including stress risers) then the same would apply.... For aluminum, however, there is no such plateau, and fatigue is still occurring even at a billion cycles.... However, if we look at the graph at 10^4 (10,000 cycles), staying below ~ 43 Ksi for aluminum, or ~ 48 Ksi for steel means fatigue should not be an issue.... The problem is that stress risers (sharp corners, drilled holes that aren't polished, etc.) can greatly increase the local stress, and are almost impossible to calculate....

bob

[lloyd-ss]

Bob,
A complicated subject, and probably best left to aircraft designers, LOL.
For steel, I have seen references to the fatigue limit being about 50% of the UTS, but it varies with alloy and the heat treat process that it has undergone, but still, generally 50%.
For aluminum, there does not seem to be such a limit, so fatigue, albeit very minute, happens whenever a load is applied and then reversed or released. But the "aircraft alloys" are much more fatigue resistant than something like 6061. I always use 2024T3 for air tubes because of the higher strength and improved fatigue resistance.
I don't know if I want to take a stand on this one, ha, ha.