+Dr. Cyders Hi Again, I was wondering where you got all the values for Se=Se'*Cload*Csize*Csurf*Ctemp*Crel the name of a book or an online link would be greatly appreciated
Chris Hoolihan These correction factors use the methods outlined in Norton's Machine Design. The same approach with similar correction factors are discussed in Shigley's Mechanical Engineering Design. I would recommend either (or both) as a good reference text for more in-depth study.
I wish we had youtube when I was going thru engineering school. Damn professors didn't even think of recording their lectures and sharing it with students.
In the pedal example the large ductile overload fracture surface as compared to the corroded fatigue fracture surface clearly shows the pedal has a high mean stress from bending load. All cracks - ductile or brittle occur and record in the time dimension - crack initiation, crack propgation and crack termination/arrest. Tensile tests are 1/4 cycle strain fatigue failures
Two question which are utterly infuriating me because I can't find anybody how can answer the questions: 1) The S-N curve for 2024-T3 Aluminium, shows stress values higher than Ftu for some stress-ratios however the number of cycles-to-failure is NOT one cycle. If a material is subjected to Ftu that should mean a failure will occur cycle 1. 2) Why can I not find any S-N curves for 2024-T42 (typical aircraft skin material) anywhere online NOR in the MMPDS?
1.) The S-N model tends to be rough approximations - without specifically seeing the curve you're referring to, the question would be whether you're looking at an extrapolated curve or actual data. Data towards the left end of the curve (low numbers of cycles) tend to be subject to variability from different sources than those at the right end of the curve, reflected in the way correction factors are typically applied. 2.) Don't know, I haven't gone digging for that heat treatment in the mmpds myself. There are guidelines for component design outside the bounds of mmpds outlined in several different technical documents.
In regard to fatigue failure, am I correct in assuming that an airplane wing would be designed for infinite life, i.e., an infinite number of cycles?Thank you for this excellent video!
+billr111 The short answer is no. Just about any machine designed should have a defined useful life. The long answer is that materials used in aircraft wings can vary, but usually includes a significant number of components made of aluminum. In reality, while models such as the approach detailed in this video are a good first shot at designing a part, for large assemblies and complex systems, it is much better practice to get real-world strain/load data from an actual physical assembly, and do fatigue testing on the actual assembly. In general, the mantra used for things like automotive and aeronautical designs usually includes this testing of real-world parts, and defining a safe life or safe load levels based on that real-world data.
When you talk about Steel having an Endurance Limit of nominally 50% UTS and Aluminium 40% of UTS; What Probability of Survival is that percentage based? If I wanted to do a design based on say a 97.7% (2 std dev) Probability of Survival; how would I modify the stated percentage values ?
Can you please elaborate why SN curve is only for high cycles of loading? and It is observed that in some text books the ordinate of SN Curve is stress range(Smax-Smin) and in some it is amplitude((Smax-Smin)/2). Since stress range > stress amplitude of applied cyclic loading, do they represent same?
Thanks you so much Doctor, I am working on piston Connecting rod assembly for 4 stroke IC engine. Does my assembly qualify for Fully reversible case? I am very confused on this, please help me. Thanks
hi, great explanation. Just 1 question. how is it that a negative mean stress is better than positive mean stress (fully reversed cycles) in terms of fatigue crack development? Isn't negative stress (assuming same amplitude) is just the reversed direction of its counterpart @ positive stress? Thanks
adam hisham Fully reversed cycles have zero mean stress, not a positive mean stress. A positive mean stress indicates an elevated level of 'constant' tension, which promotes crack growth, leading to a lower life at a given alternating stress, or, looking at it another way, a lower allowable alternating stress for a given part life in cycles. See my video on the Goodman Diagram (ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-fei7iQd_lVE.html). A negative (compressive) mean stress doesn't generally cause this, as it doesn't promote crack growth the way a tensile stress does, but the alternating stresses can still cause fatigue. If you have a copy of Norton's or Shigley's machine design texts, they both discuss the phenomenon. Does that answer your question?
Dr. Cyders Owh. I'm getting the jist of it but still a bit lost. Lets say there a beam. A load is applied on it and the beam bends to a certain degree. Then the load is reduced so that the beam 'unbends' itself towards its initial/resting position. Before the beam reaches resting position, the same load as earlier is again applied. The cycle continues. In this situation we could say that a positive mean stress is taken by the beam right? Now lets apply the same load conditions to the beam with a change only in the direction of the bending of the beam. Is the beam now experiencing negative mean stress?
Your explanation would actually include points that would experience a negative [compressive] mean stress and points that would experience a positive [tensile] mean stress (assuming you were talking about a simple cantilevered beam). For clarity's sake, say we have a uniform cantilevered beam, and you stand on the free end of it. You have statically loaded it, and it has a maximum stress at the rigid wall, tensile at the top of the beam, and compressive at the bottom. Now say you jump very lightly up and down on it - a simple model would be that the force on the beam goes to zero when you're in mid-air, and up to your body weight when you're on the beam. This would be a repeated load. In this case, the state of stress at the top of the beam at the wall would oscillate between zero and some finite positive value, so you would have a positive mean stress at that point. At the same time, the stress on the point at the bottom of the beam at the wall would be oscillating between zero and a compressive stress; therefore, that point would have a negative mean stress. Thus, if we jumped up and down enough to fatigue the beam, the fatigue crack would likely begin at the top. To continue the example, we could counter this - we could put an axial compressive stress on the beam, putting the whole thing in a constant state of compression, say, by running a screw through the whole thing and tightening it. With enough pre-compression, we could move even the mean stress on the top of the beam to zero or below, and we could therefore improve the life of the beam without changing the beam's geometry or load.
Sometimes you can find fatigue limit data, sometimes you can't. It's usually a bit harder to find than straight monotonic data. If you have access to a solid library, there are a number of fatigue data books out there.
+QuantumTeapot There are several methods to deal with multiaxial stresses, probably the simplest of which is to a.) assume the worst case that the stresses will sync up at some point, b.) create von Mises effective stresses for the alternating and mean components of the various stresses on your worst element, and c.) use the resulting alternating and mean effective stresses on your S-N curve and Goodman diagram to assess the projected lifetime and factor of safety.
