Could you do an explanation in a future post of the differences between a part (say, a bracket on a car) failing due to a migrating stress fracture versus a total, quick failure where it just breaks in two? Is that the same stress dynamic in play with different outcomes, or are they different processes entirely?
@Alex: Usually, a crack which grows over time is due to fatigue, which I'll cover in my next installment. (In plastic parts, cracks which grow over time could also be due to environmental stress cracking, which I've written about before. In metals, there is a phenomenon called stress corrosion cracking, which is analagous to environmental stress cracking; I might write about this later).
Based on what I've seen in my career, fatigue failures are actually far more common than overload failures. However, overload failures are the easiest to understand, which is why I wanted to cover them first.
The mechanics of fatigue are a little more complicated. As I'll discuss, a common mistake is to treat "fatigue strength" as though it's a property like yield strength or ultimate tensile strength. It's not. But the big picture is the same: you need to understand the forces that act on the part, and the properties of the material from which it is made -- as well as all of the variables which might cause either one to vary from its normal value.
Dave: In the kinds of parts that are mentioned here, such as the brake cam, are the parts typically designed in accordance with the yield strength of the material, or is there some "allowable stress" design method that's set forth that is not dependent on yield? If yield is not used as criteria, does it make any difference in terms of failure rates?
@Chuck: Other engineers should feel free to weigh in on this, but in my experience, it's most common to design to the yield strength, with an appropriate factor of safety. Doing this should protect you against overload failures, provided that (as I pointed out in the article) the loads are what you think they are, and the yield strength is what you think it is.
Dave, I think the weight is eventually distributed across the area, and then it can bear more weight than concentrate to particular points. I think in most of the industrial wing, the stress tests are doing for a mass areas rather than stress test in cubic/cm sqd.
@Mydesign: You're right that loads redistribute to a certain extent as a result of localized yielding, so that a linear FEA which predicts a stress greater than the yield strength in a small region doesn't necessarily indicate failure of the component. This is why designing to "get the red out" of a FEA model, without any insight into the physical situation, can result in overdesign. On the other hand, stresses below the yield strength can lead to fatigue failure if they are repeatedly applied. To get a handle on fatigue, it's important to know how the loads on a part vary over time. This is what I will discuss in the next installment.
The subject of "getting the red out" is my current challenge as my design department is just now using FEA. I have had other FEA users at other company locations run FEA on parts in the past & early-on found this issue on linear static analysis of steel forgings. The loading is static and the cross sections are "L shaped" so that there is a bending stress at the inside transion from vertical to horizontal. I've found no amount of thickening the cross section ever completely eliminates the red (below yield). I have seen a report from a P.E. on a similar part where the P.E. concluded that "slight yielding in this area relieves the high stress concentration and then distributes the load more evenly accross the cross section of the part" - or words to that effect.
That's all well & good but my issue is how to justify the "remaining red" in a report that will be reviewed by other engineers who are not M.E.'s (Petroleum Engineers) - and who are the customer in this case. Any brief suggestions and or recommended literature on this subject would be most welcome.
@bentarrow: I assume you're using a linear FEA package. Nonlinear FEA would give you a much more accurate picture of localized yielding and load redistribution. Otherwise, you may just want to point out to your customers that the high stresses at the transition are a result of the assumptions of the FEA model. This may or may not be a satisfying answer to them, but it's true.
If you can do actual physical testing, this might also help to convince them. You could look into using a product like Stresscoat to measure the actual tensile strains, and compare this to the results of your FEA model. If the actual part is too big to test in the lab, you might be able to use a scale model; FEA can help you understand how to appropriately scale the loads so that you get the correct stresses.
Thanks, Dave. I brought the topic of yield strength because I seem to remember something from my distant past called, "working stress" design. I believe working stress was not based on yield strength, but I'm really not sure.
@Chuck: Usually, in mechanical engineering, the term "working stress" simply means the stress which a component experiences under working loads. Typically, engineers and designers try to keep the working stress below the yield strength, divided by an appropriate factor of safety. (The yield strength divided by the factor of safety is sometimes called the "allowable stress" or the "design stress").
Hey @Dave Palmer, thanks for an awesome post! It's been a while since I've left my industry position, but all of the terms came flooding back. My development team was tasked with creating measurement techniques to complement / verify the FEA. We developed Temperature-sensitive Paint to measure and track propagating fatigue cracks while the part was under test. We developed Pressure-sensitive paint to measure aerodynamics and surface stress distributions. One of the most interesting projects was developing Strain-sensitive Paint for Ford Visteon to visualize whole-body strain-distribution on truck axles (now commercially known as Strain-Sensitive Skin, S3). All of that development was done in the late 1990's. I'm not sure how popular the techniques are now...
As mentioned earlier, repeated stresses/cycles on an assembly are a major contributor to fatigue and crack propagation. The biggest contributor to repeated cycles is vibration. Sometimes it's difficult to observe, but even high frequency vibration (though very small displacements) can be a fatigue factor due to their high cycle rate.
Vibration can be an issue when attaching a component to a moving machine (frequency depends on the machine dynamics), when designed to handle siesmic vibrations near fault lines (relatively low frequency <10Hz), or just designing to handle transportation to the end user (between 2-500Hz). The frequency and amplitudes vary, but the main goal is to design components with resonant frequencies well above what the sample will see while in use or transport, and when designing machinery, to avoid stacking resonant frequencies so the components aren't exciting each other's resonant frequencies while in use.
My experience has been that operators have the unique ability to find every unintented use of a piece of machinery - causing real eningeering challanges when it comes time to find out what REALLY went wrong.
It's interesting to watch what happens when a part is redesigned, "beefed up" because it's been breaking in the field. If the redesign works correctly, the part is no longer the "weakest link", and something else now is.
The usual progression is a series of parts end up being redesigned, one at a time, as each becomes the weakest link in turn.
@TJ McDermott: The all-important question is, "Where is the load coming from?" As you point out, responding to failures by simply beefing up whichever part happens to break often just results in other parts breaking. At the end of the day, it results in assemblies which are extremely robust, but which are also unnecessarily heavy and expensive.
There's no substitute for a solid engineering analysis of the mechanical system as a whole. If you understand where the load is coming from, you can address the source of the problem, rather than constantly beefing up parts to compensate.
As I said before, the part which breaks is often an innocent bystander. Don't blame the part for the inadequacies of the design!
Dave Palmer certainly got it right in the comments about nonlinear stress analysis. And I would point out that vibration is often an unanticipated mechanism for the excess stress that starts those cracks. In addition, the vibration often leads to the fatigue failure that starts the process. Of course, the amplitude of the vibration is much greater in cases where there is resonance. So now there is a whole list of things to beware of.
The 100% solar-powered airplane Solar Impulse 2 is prepping for its upcoming flight, becoming the first plane to fly around the world without using fuel. It's able to do so because of above-average performance by all of the technologies that go into it, especially materials.
With major product releases coming from big names like Sony, Microsoft, and Samsung, and big investments by companies like Facebook, 2015 could be the year that virtual reality (VR) and augmented reality (AR) finally pop. Here's take a look back at some of the technologies that got us here (for better and worse).
Good engineering designs are those that work in the real world; bad designs are those that don’t. If we agree to set our egos aside and let the real world be our guide, we can resolve nearly any disagreement.
Focus on Fundamentals consists of 45-minute on-line classes that cover a host of technologies. You learn without leaving the comfort of your desk. All classes are taught by subject-matter experts and all are archived. So if you can't attend live, attend at your convenience.