0:00 reading quiz and review of Griffith fracture toughness 6:10 learning objectives 6:40 ceramic fracture, stress corrosion, static fatigue 10:23 polymer fracture 15:21 testing for fracture energy (impact testing) 16:37 ductile - brittle transition 19:34 cyclic stresses and fatigue failure, fatigue limit 27:35 how do we deal with variation in failure with S-N curves? 31:29 frequency dependence of S-N curves 33:51 mechanisms of cyclic crack growth 36:50 Paris law for crack growth 38:54 making the exponential crack growth equation look linear using log axes 40:29 solving the crack growth equation by integrating from initial to final crack length 41:57 determining critical flaw size 45:28 worked example of crack growth problem for number of cycles til failure and largest tolerable flaw size.
Okay. Welcome to class. Welcome to your Wednesday. Okay. Let's start by doing a reading quiz.
We didn't do it on Monday because it's the day after the exam. So a lot of these questions
are from the reading, but we covered them on Monday already.
In a crack splits through brains, this is called transgranular cleavage, and it is a
hallmark of ductal fracture or brittle fracture.
Okay. Answers in transgranular. Oh, that's supposed to say transgranular. That's a typo.
If it helps at all, when it is transgranular and it splits through the grains, that makes
it a more flat surface. So is that ductal or brittle? Answers in. Okay. We're going
to wrap up this poll. So if it's a flatter surface, that is definitely brittle. Okay.
Let's do the next one, which is fracture is spontaneous, meaning that the crack growth
is unstable, right? It's going to happen on its own without having to mess with it. When
what two energy terms are equal to one another? Free energy and surface energy, surface energy
and chemical energy, strain energy and free energy, or strain energy and surface energy?
Okay. Wrapping up the poll. Is it A, B, C, or D? If you don't know, phone a friend. We're
going to end right now. Okay. Definitely the last one. Okay. One last one. What is the
main difference between ductal and brittle fracture? A, ductal fracture absorbs a lot
of energy and has a significant amount of plastic deformation. B, brittle fracture absorbs
a lot of energy and has a significant amount of plastic deformation. Or C, ductal fracture
is not associated with creep, but brittle fracture is.
Okay. People think, everyone thinks A. Everyone got the right answer. That's amazing. Has
this happened before? Well done. Well done. Okay. So last time we left off, we did our
first calculation. Our first calculation, what we call Griffith fracture toughness,
okay? Griffith fracture toughness, we're going to come back to it over and over. I guarantee
there will be a question that uses it on the midterm, the next midterm. There always is.
And the key to using the Griffith fracture toughness equation is the following. You've
got the left-hand side of your equation and you've got the right-hand side of your equation.
The left-hand side of your equation is a fundamental material property, right? So you look up some
alloy or some material or whatever, it has just a fundamental property. On the right-hand
side of your equation are things that can change based on your experimental conditions.
For example, the crack size and the loading, as well as the crack geometry. If at any point
the right-hand side of this is greater than the left-hand side, it will fail in brittle
fraction, right? It will spontaneously crack. Okay? That's as simple as it is. If the right-hand
side of your equation is larger than your fracture toughness, it will fail, right? Or
in other words, if you know what the fracture toughness value is, like we did in this problem,
we told you it's 60.4 megapascal root meters, then you can figure out what will make it
fail by making the left-hand side equal the right-hand side, right? In this case, we knew
what load we were going to apply and we asked the question, how big can the flaw be, right?
We could have done the opposite. We could have said, all right, there's a flaw in your
material that's five millimeters big. What's the highest load that this component can sustain,
right? And you would solve for sigma in that case, right? So there's lots and lots of ways
to ask questions around this because there's multiple variables that we could choose to
ask about, okay? We'll do another example of it later today. Today, our learning objectives
are first, identify what is static fatigue and talk about how that relates to failure
of ceramics in particular. We'll describe crazing. We'll talk about impact testing and
ductile to brittle transitions. We'll use SN curves to observe what's called the fatigue
limit, fatigue like things get tired, materials get tired when you use them over and over,
similar concept. And then we'll probably only have time to do Paris's Law to predict crack
growth. But if we do have time, we'll get into creep of materials, right? So let's start
with static fatigue. When ceramics break, as everyone got right on the quicker question,
there is no plastic deformation. Again, we gave the example of your dinner plate. You
can't flex that at all. Like a piece of chalk is ceramic. You can't flex this at all. It'll
just break if I keep trying to bend it, right? Whereas other things like a rubber band, you
can deform quite a bit and it doesn't break, right? The stress raisers are typically pores
or impurities. And because these things can't accommodate any plastic deformation, they
tend to have really low fracture toughness values, the lowest of pretty much everything,
some polymers really, right? So when we say low, what's a low value? Well, they're typically
less than 10 megapascal root meters. That's low for most ceramics. Whereas metals like
steels and aluminums can be 20 or 80 or different values. And polymers can be kind of in between,
actually can be around ceramics, okay? Something called slow crack growth is possible, right?
So here's our Griffith fracture toughness equation, right? And I said if at any time
the right-hand side of that equation is larger than the left, or once it's equal to it, this
thing will fracture. So something that can happen is this. Initially, your crack size
is small, right? But as you cycle this thing in use, maybe you're pulling on it and compressing
and you're going back and forth. If you do that over and over, what can happen is your
crack can grow over time, right? Cracks don't have to stay the same size. They can steadily
grow. And as it grows larger, it can reach a critical size, right? It can reach a...
Today's going to be one of those days where OneNote ruins our life. It can reach a critical
size, right? Definitely going to be one of those days. Let me stop this and restart it
and hope that that fixes things. I think it's having a phantom touch issue on the screen.
I need to get fixed. Okay. So if that thing grows, it can grow large enough that you will
actually fail, right? So we need to determine what will make it grow. One thing that can
make it grow is when you're not even cycling your stress, it can still fail. This is called
static fatigue, meaning you apply a load to it and that load causes it to fail, even though
that load is static. What's typically happening is at the tip of the crack, the crack is growing.
If a little bit of water, for example, gets at the tip of the crack, it can dissolve the
material at the tip of that crack. Now, normally you don't think of a ceramic as being particularly
soluble in water, but when it's a really small radius of curvature, so therefore a
high curvature, then your solubility can be much higher. So the very tip of that crack
actually can be dissolving and that makes your crack continue to grow, right? That's
one possibility. Because of that, they'll call this stress corrosion cracking sometimes.
And if you have a ceramic that's going to be in liquid, they can be particularly susceptible
to this. But basically, again, you're dissolving away your material. So it starts out in some
crack length, and as you slowly dissolve it, that crack just gets longer and longer
and longer, okay? Eventually, it's going to be long enough that it satisfies the critical
flaw size and the single fracture. You can see this, by the way, if you scribe a piece
of quartz tubing and you try and break it, it's pretty hard to break. That stuff's pretty
strong. But if you just dip that under water after you scribe it, now you've got water
at the tip of that crack, you can pop it at least half the strength. It's much easier
to break. All right, let's talk about polymer. When polymers fracture, typically the yield
strength, these things are not as strong as metals or ceramics, but they can deform a
lot more, right? You have thermosets and thermoplastics. We'll talk about those in a couple of chapters.
What's the difference? Yeah, question? Great question. We're going to come back to it.
Let's talk about, that's the point where it starts to deform. The yield strength starts
to yield. We will come back to that in a few chapters. But yeah, for now, think of it as
you're going to apply a load until it starts to bend or something. Yeah, question? Yeah.
It's permanent or plastic deformation as opposed to elastic. So rubber band, you take a rubber
band and stretch it. We say rubber band, you call it an elastic band, right? Because it's
elastic and you stretch it and it goes back to where it started. If you pull it too far,
you now enter the plastic regime where it's permanent. It's non-reversible. There's two
different types of plastics. One tends to be more brittle. One tends to be more ductile.
We'll come back to this. You don't need to know this just yet. And you've probably seen
this. Like an old, like imagine like a fifties era telephone. That's a plastic, but it's
not very forgiving one. If you drop that, it could actually break into pieces. It's
kind of brittle. Whereas a lot of plastic, like a geoguy, if I bend them in half, it's
not going to break. It's going to bend, right? Of course, it does depend on two things. It
depends on temperature and it depends on the strain rate. That's the critical thing. Remember
the banana? The banana could be very, very ductile or it could be actually pretty brittle
depending on whether you break it fast or if you break it at a higher or low temperature.
So temperature and strain rate are critical for determining ductile versus brittle. And
then you can also get something called crazing occurring. Crazing is when you get these regions
of localized yielding that connect the crack, right? See these little bridges happening
in this thing? You see those little bridges forming? That's called crazing. Do you think
that that increases or decreases the fracture toughness? It should increase it. How would
you guess that? Well, what are our two rules of thumb? To make a material tougher, you
have to do one of two things. You either have to, you ready? Tell your neighbor what you
think the two things were. To make a material tougher, what can you do? Okay, let's have,
let me ask David in the Alaska shirt. What is one thing that you could do to make a material
tougher? Okay, how would it make it tougher? What's the mechanism that's happening? Okay,
maybe you guys want to add on? Yeah. Yeah, remember the two rules of thumb. If you make
it harder for the crack to propagate, like a secondary phase like you added, maybe it
hits that secondary phase and has to go around it or something. Or, yeah, question Rosa?
Yeah, because for that crack to continue delaminating, it has to pull out those bridges basically.
So if you make a crack harder to extend or if you make it harder for a crack to start,
either of those things will increase the toughness of the material. And another way to think
about toughness is how much energy gets dissipated as the crack moves forward. In this case,
if you've got a bunch of little bridges in there, that's dissipating a lot of energy,
right? You have to put energy into that. So it's storing more energy as the crack grows.
So you made a tougher material, okay? So crazing is good and it's why polymers don't
have a lower fracture toughness, right? Because they're typically characterized by having
crazing occur. Yeah, Spence? Let's wait until we get to the polymers chapter. We'll come
back to it. So we will return to polymers. Oh, that was a clicker question. We already
did that one. All right, get your free clicker question. Did crazing increase or decrease
fracture toughness? Just did this, so everyone get your answers in. Yeah, they don't have
to. I'll show you some examples later on. You could have like a laminate structure,
like a bunch of plies laid up. And when you try and pull that apart, you'll still see
some bridging like that. So they can occur both ways. And it depends. If you're talking
about a composite, you can engineer it to be one way or the other actually. Okay, so
answers in, 100%. Will it happen again? Oh, sorry. Okay. All right, how do you test fracture
toughness, right? Fracture toughness is, again, it's this quantitative material property that
tells you how resistant to fracture something is. So how do you test that? We'll talk about
fracture toughness in terms of energy in a few chapters. But one thing you can do is
you can look at how much energy gets absorbed before versus after, right? So you take a
pendulum. Mechanical engineers, have you guys used one of these before? You haven't used
one of these fracture toughness? You start it out and you note the position that this
pendulum is at. So it's a pretty heavy weight. Your sample is sitting down here at the bottom.
And your sample has a little notch in the middle so that it breaks in a specific spot,
right? And this hammer basically comes swinging in, right? It comes swinging in right here
and it hits your sample. And when it hits it, it's going to break it. You know, if it's
brittle, it'll break it. And then that hammer is going to keep on swinging, but it won't
go up as high as it started, right? There's going to be a difference in the height of
your starting point and the height of your ending point because some energy got transmitted
to your sample, right? And so this is a tool that basically measures the energy as you
create these new surfaces and as you strain the material, right? So it's a measure of
fracture toughness. So they do this. They'll do these Sharpie notch testing is what they
call them. Sharpie pendulum or Sharpie notch tests, right? They'll do them at different
temperatures and you often see this behavior. Some materials will exhibit what's called
a ductile to brittle transition. So this mild steel, for example. Now, which side is ductile
and which side is brittle? Turn to your neighbor and take a guess which one should be which.
Like here's clearly the transition temperature somewhere around here. Which side is which?
Okay, let me pick on Ari. What do you think? So she thinks that the higher side, so on
the right is ductile. What's your rationale? It's just absorbing more energy. If you're
absorbing more energy, you must have made it harder for that crack to travel through.
You must have made it tougher material. And almost always, in fact, I don't know of an
exception. You have ductile materials at high temperatures and if it becomes brittle, it
would be at low temperatures, not at higher temperatures, right? So when you cool things
down, they become more ductile. I don't think vice versa, right? But not all materials have
this behavior. Some, it's really remarkable. So some type of steels, this is a big deal
because imagine that you're in World War II, right? Or Takara, question? Great question.
Your ductile transition could be associated with a phase change. I'm talking about with
a single phase that increasing it doesn't make it more brittle. But you're right. If
you go from a low temperature phase to a high temperature different phase, that could be
more brittle. So that's actually a really good observation. If you're in World War II,
which we were, right? And you were making boats to go into the Baltic, which we were,
they tried to make them as quickly as possible because our Navy got decimated, right? And
we basically had to start from scratch. So we made boats as quickly and cheaply as possible
to try and get up to speed in the war. And we made what were called liberty ships. Liberty
ships because we made them quick and they were going to spread liberty or at least prevent
us from moving forward, right? The problem is they used a steel that when you go in the
cold ocean water, like up in the Baltic Sea, you see a ductile to brittle fracture, right?
And a ton of these boats actually split in half before they ever left the harbor where
they were being worked on. As soon as they put them in the water, a lot of these things
actually split right in half because they failed in a brittle fashion as opposed to
being ductile. Yeah. You'd have to ask your mechanical engineer colleagues. I am at a
loss to explain some of the other details. Some people think this is why, well, some
people thought this is why the Atlantic, sorry, the Atlantic, the Titanic. Yeah. The Titanic
sunk. They said, oh, it was way up north where it's brittle. It hit an iceberg. And instead
of just denting, it actually just fractured. And that's why it went wrong. There's now
been studies, even really clever material science studies, actually looked at the rivets
and it was actually really poor quality rivets that they used that actually sheared and then
plates actually probably just fell off. So it probably wasn't ductile brittle transition
fail there, but it definitely was with the Liberty ships in World War II. So this is
a big deal. Yeah, World War II. So let's talk about our next concept, which is fatigue,
right? Fitting for a Wednesday afternoon fatigue, fatigue with materials when they get tired,
right? You're using them. Maybe you're doing a cyclic stress over and over. And after a
while, it's just like, can't do it anymore. I've had enough of this stress nonsense. I'm
just going to fracture, right? That's basically what's happening. You have some sort of loading
condition. It can be a really carefully controlled and repeatable loading condition like right
here where the average stress is zero. And it in a very predictable way goes up and down
and up and down, right? In a sort of sinusoidal function, or it could be an offset sinusoidal
where it's always under tension, but that just goes between high and low tension, right?
This, again, is going between compression and tension, right? Or it could just be random,
which is the more common scenario, right? Say like temperature, right? Every day, we
know that it gets warm in the day and it gets cold at night, but the actual amounts vary
quite a bit, whether the pressure system around, whether it's storming, right? So it might
look something like this. If you look at the thermal stresses that that temperature fluctuation
causes. So most materials are experiencing cyclic stress conditions, whether you're designing
them to or not. Just the fact that we live on a planet where temperature changes means
everything's going to have some sort of cyclic stress. Doesn't mean that it's big, but it
means there's something. So we need to take into account what this does to material. Basically,
you can calculate the median stress. So your average stress is easily, it's just your max
plus your min divided by two, right? So your median stress is not zero in this case, but
it would be zero in this case, right? Here's your median stress there. You can also calculate
the stress range, which is the distance between your maximum and your minimum, right? So that's
your range. And half of the range is what we call the amplitude, sigma A. The sigma A,
the stress amplitudes, is what we usually use when predicting fatigue from SN curves.
Let me show you what that looks like. Yeah, question? Ah, great. We will talk about that
in just a moment. Yeah, great question. Basically, when can you ignore the compressive part?
Yeah, we'll come back to that. So first off, if you were going to, you guys are all engineers,
you're going to work for companies, and maybe your boss says, hey, we need to test the life
cycle of this thing, right? So put it through a million cycles of compression and tension.
What are you going to do? What does that actually look like as a test? How would you do that?
One way to do it is a cheesy way. If you walk through Ikea, you've seen the robot closing
doors over and over, opening and shutting things. That's one way, right? How else could
you do it? That's going to take a long time. Yeah. That's a great way. Later this chapter,
we do something actually a lot like that. It's called the Larson-Miller parameter.
We'll probably get to it on Friday, not today. One way to do it is to do, they take a rod,
right? And they rotate this thing, right? But it's not at a perfect zero angle.
It's actually at a slight angle like that. So your rod is actually doing this thing.
And this is like very, very exaggerated, right? It's only maybe like a half a degree,
whereas this is 10 or something. So if it's at a slight angle, you rotate it.
Imagine what's happening. On the top, that point is currently under compression as I
rotate it, right? Because it's sort of bent in like a V. But when that rotates to the
bottom, now it's actually under tension, right? Tension and compression. And then you can
just rotate that thing over and over and over and over and over. And eventually it's going
to cause a flaw to form at the surface, which will propagate through. So this is one way
to do these lifecycle tests. I think it's one of the more common ways. There's other
ways, but this is one of the common ways. So you do that until it fails. And you keep
track of how many rotations it did until it fails. And you generate a curve that looks
like this. On the y-axis, you're plotting your stress amplitude, right? And on the x-axis,
the number of cycles until it failed. And you can then overlay all sorts of different
materials, right? So you've got steel, which exhibits what we call a fatigue limit, and
aluminum, which doesn't really. So what is a fatigue limit? If you take a paperclip,
which is aluminum. Yeah, question or comment? It's a type of bending test. It's a bending
test that's alternating. It's like an alternating bending test, right? Because if you took this
thing, and if I were to start bending it, the top is under tension, the bottom under
compression. And if I switched it, this is basically automating that process by spinning
it, right? So you see two very different behaviors. The steel, it reaches what's called this fatigue
limit, but aluminum doesn't. If I had an aluminum paperclip, which I always forget to bring
If I start to bend it back and forth, by the time I do maybe 15 of those, it will snap
in half. If I had steel, it wouldn't. First off, it's way harder to bend, but it won't
actually snap, right? As long as I stay below a certain threshold. Meaning steel, if I stay
below, this looks like 300 meganewtons per meter, that's a megapascal. If I stay below
300 megapascals with a steel paperclip, it will never break. But in aluminum, if I just
keep on stretching that, or flexing it, eventually I'm going to cross below wherever my threshold,
maybe I was like flexing it right here. I'll cross below it, and at that point, I'll have
done a million cycles, and it will fail, right? So this is pretty interesting. As an engineer,
we care about this, because you can choose what materials you work with. And if you know
that this thing has to survive a gazillion cycles, you don't want to pick something like
aluminum, because it's going to get weaker and weaker and weaker, and it's eventually
definitely going to fail. You'd like to pick something like a steel that has a fatigue
limit. What causes a material to have a fatigue limit, we can't answer just yet. In a few
chapters, we'll be able to come back to it and talk about some reasons why some materials
have a fatigue limit and some don't. So we'll come back to that. Any questions about SN
cycles in these plots? So if I said, I need aluminum to last 10 million cycles, you would
say 10 million is right there, right? Therefore, I better not load it to more than 175 megapascals
at the most. Now, what's the problem with this? Do all materials, when you order a batch
of 20 bars of aluminum, and you were to bend them or stretch them until they break, will
they all break at the same stress? Definitely not. There's always variation, right? It might
be really close variation, where there's only like plus or minus a few megapascals, or it
might be plus or minus 50 megapascals, right? So how do we deal with that? What do you think?
Zach, you got a comment? Well, stress is the force over your area, right? So let's
say you've got like a cylinder, and you're pulling on it from the top and bottom. So
you're applying a load. So your cross-sectional area, if you make it wider, you go to a smaller
stress. And so you could design your component. If you know that this thing will break, if
you go above 175 megapascals, you can just make your component dimensions larger, such
that the stress on it is under this value. And typically, we do that. And then, because
we're engineers, we say like, oh, and put a safety factor of two or three on it. And
you say, we're actually going to make it so thick that it should never experience more
than 50 megapascals, right? In a few chapters, we'll talk about Weibull statistics, which
is a way better way to go about it. Rosa? No, they're based on actual tests, where they
actually took a rod, they bent it slightly, you know, something like a half a degree or
something. They spun it until they observed some sort of failure, whether that's it deformed
over a certain amount or actually fractured. It depends on whether it's broken, ductile
or brittle. But no, these are experimentally calibrated. And then to answer the question,
maybe how about this? Turn to a neighbor and tell them, how do you deal with the issue
of the fact that samples don't all break at the exact same yield stress? They break over
range. What do you need to do to your S versus N curve to account for that?
Okay, anybody have a thought? What on earth do we do? How do we account for variability
and data with these S? Again, this is called an S-N curve, stress versus number of cycles.
Anybody have a thought? Or did anybody's neighbor have a great thought? I'm going to
sell them out. Yeah, Kyle had a great thought. What would you do, though? You've got this
curve. What do you think this is, first off? Yeah, maybe that's the first question. What
do you think this represents? Probably, okay, best fit, or in other words, like an average.
So like this might be where 50% failure occurs, right? 50% of your samples is right there.
So now how would you account for, if you need to know where not even one could fail,
what do you need to do? Anybody have a thought? Yeah, Matt. Lots and lots of testing. You're
exactly right. You would have to introduce other lines. You'd have to have another line
right here. Would this be corresponding to a higher or lower survival rate, as opposed
to this one? This one is a higher survival rate. This might be like 1% of samples fail,
right? Because say, after a number of cycles, let's draw it the other way. If you're going
up from a certain number of cycles to failure, some of them will break at a really low value,
but not very many, right? Some of those will break down here at stress low, and some more
will break at a higher stress. So actually, what you really need is a bunch of curves
that represent different failure rates, right? 1%, 25%, 50. So if you find a good provider
of steel or aluminum or whatever, they're going to provide a curve that has various
different failure rates, not just one. Because one is just an average, and you don't know
what the distribution is from that plot, so you're left to sort of wing it. This is a
much more quantitative, better way to talk about failure and SN curves. Can I answer
any questions about this? We can move on. Yeah, questions? Well, let's say that you
got this plot, and this is what they gave you. And you know, from what your boss told
you, that you cannot have more than 1%, you know, 1 in 100 fail. What would you do? You
know that it has to survive this number of many cycles, right? NF, whatever that is,
number of cycles to failure. Maybe that's a million, maybe it's 100, right? You know
that you can't design above that stress. You'd stay under it, yeah. Even though you know
that's only when one's going to fail, most of them won't. It depends on what your
specifications are, right? When you sell a product, if you sell an iPhone, if 1 in 100
are defective, you're not going to do very well, right? If you're selling chalk, and
you open up the box of chalk, and 1 in 100 are broken, you're lucky, right? So it just
depends on the application, right? Yeah, hand over here, Anthony?
So basically this graph is showing just when these materials will fail.
At different rates. So this is all for aluminum, or whatever your material is, steel, whatever
it is, right? But at different failure rates, okay? Okay, let's keep going. Some materials
like polymers can have a really strong dependence on frequency, right? So you're rotating this
thing, right? Again, I've got this pencil, and I'm going to bend it a little bit, and
then I'm going to rotate it, right? So I'm constantly alternating that stress. Some materials,
if you rotate this really fast, you'll get a different S-curve than if you rotate it
at a slower speed. Why do you think that might be the case? Turn to your neighbor and discuss.
Okay, what do people think? Why is it that polymers, it depends on the speed at which
you spin it, whereas metals, less so. Any thoughts? Anybody want to berate a guess?
Is it John? Yeah. It's probably wrong. It absolutely is. You're definitely not wrong.
It has to do with heat. When you stress something, right, you're actually causing atoms to shift
around and move, right? You're generating through internal stresses quite a bit of heat.
If you were to touch that thing after it's been spinning for a while, if it's been spinning
fast, it's going to feel much warmer than if it's been spinning cold. Now for metals
and ceramics that melt at bonkers temperatures, it doesn't matter. For plastic that might
melt at 100 degrees, that could be a significant change, and it's going to change your elastic
constant. So yeah, that's really why. So it's the heat generated can become a non-trivial
compared to melting point. Okay.
Like in singing to a glass, sometimes if you hit the right frequency, it can shatter.
Okay.
You also have to do when you spin it, it also causes micro-vibrations.
Absolutely.
So it has to do with just whatever is at 100.
Absolutely. I have never even thought about that, but that's absolutely right. Yeah, the
resonant frequency, you could be rotating at something similar to that, which could
cause additional stress multipliers. Okay? All right. So fatigue, when you break something
due to fatigue, again, you're cycling it over and over and over, you have to have two steps.
You have to start the crack and you have to grow the crack. So the total number of
cycles is going to be Ni plus Np, right? Initiation plus propagation. Stage one, cracks initially
propagate slowly on something called a slip system, which we will come back to. We haven't
talked about that as yet. And then they grow, and they don't grow at a constant rate. They
tend to grow in an exponential fashion. I'll show you what that means. Physically, what's
going on is let's say this is your crack initially, you pull under tension and you actually stretch
that crack a little bit. And then when you compress it, you leave behind, you're additionally
one step further than when you started. So it's essentially only growing under tension
is what that is. Under compression, it's just closing the crack. It's not usually growing
it very much. It's usually only growing under tension. So Rose's question earlier was a
really good one. She said, okay, if you're cycling it between, say, positive 100 megapascals
and negative 20, what's the range in stress? It's actually only between 100 and zero, because
the negative part doesn't actually grow the crack. Anytime the stress is positive, we
call that tension stress, it's going to grow over that entire range. But when it's negative,
in this class, we're going to assume that it's not growing. That's an assumption. It's
not valid for all materials. It's a pretty good assumption. Yeah. That's all materials.
For all materials, we're going to assume in this class that they only grow under tension.
Again, not always the case. Pretty good for this class. It's a pretty good approximation.
Because of that, when you look at things that break after fatigue, you see these, they'll
call them clamshell patterns, striations, benchmarks. It's all talking about the same
thing. You're seeing a bunch of little lines, and those represent an individual cycle. Maybe
this was the axle of a car, and eventually it broke. You're seeing individual rotations.
If it's maybe day-to-night sequences, you're seeing where the crack incrementally moves
forward a little bit. If you look at it, the cracks are closer together, and then they
get further apart. They tend to not be constant. The crack tends to grow faster at the end.
In fact, you get something like this. Here's a plot of crack length versus number of cycles
for three different stresses. Which one corresponds to which? Which one's your highest stress?
Farthest to the right? Think about that. For 500,000 samples, you grew your crack to here.
In this case, after 500 samples, you would have grown to infinite. It's not even on the
graph. Actually, this must be your stress high, and then it's getting lower. The stress
that you apply will change the crack growth. Generally, regardless of what stress you choose,
you're seeing what we call exponential behavior. It looks like a power law. It's taking off.
Even though the crack, after the first 100,000 cycles for all of these scenarios, hasn't
grown that much, the next 100,000 really can take off, depending on your stresses. It's
exponential. We can model that with this awesome equation, which they took out of the
third edition, because I think you're afraid of calculus. But we're not. We're not afraid
of calculus. We can do this. This is a differential equation. You've got dAdn. That's the change
in the crack length with respect to the number of cycles, dAdn. After one cycle, how much
further along is your crack? That's what that's basically predicting. It's going to be equal
to A. Sometimes they use C. I apologize. As if this field isn't hard enough, they go
and use multiple symbols. A or C. Then it's delta K. That's the range in your stress intensity
raised to the exponent n. A and n are materials constants. We call those the crack growth
constants. For every material, you could look those up, and people know them. Then, delta
K is just your Kmax minus your Kmin, and K is just equal to Y times the stress times
this square root of pi thing. All we're doing is we're taking the range in your stress.
In this class, we're only considering positive stresses. If the stress ranges from positive
500, negative 250, your stress range is just 500. If it ranges from positive 300 and positive
100, then your delta stress is just 200. We only care about the positive stress because
that's when it's under tension. How do we do this? Typical failure due to crack propagation
plots look like this. You've got this initial region, region 1. That's where your crack
starts to grow. Eventually, if you plot this on a log plot, this is a log scale over here,
and this is a log scale here, then you get this linear region. The reason it's linear
is because your equation looked like this. dA dn equals A to the delta K to the n. Question,
Rob?
Yeah, so sorry, Rob. In that formula, what was the exponent n?
What was what now?
The exponent n.
Exponential n and capital A, those are both constants that you just need to know in order
to know how the rate of a crack growing in your material. Those are things that I would
either give you or ask you to solve for by giving you a plot like this. Now, why does
it look linear? Why does this equation look linear when you plot it on a log axis? Well,
let's take the natural log of both sides. We now have natural log of dA dn equals the
natural log of A plus the natural log of delta K, and then this is, sorry, let me make a
little space, n times that. A property of natural logs is if you take two natural logs
that are being multiplied by each other, you can add those, right? That's why we put an
add there. Yeah, Kara?
That is n times ln K. That's what it's supposed to be. Let me rewrite that clearer. We brought
the n down from an exponent because that's what you can do with natural logs. n becomes
multiplied by natural log of delta K, okay? Or in other words, if you plot the natural
log of this, which is our y, and you plot it against the natural log of your stress
intensity range right there, then that should be your slope, right? This should be x, and
this whole thing should be your y-intercept, right? So it just becomes y equals mx plus
b. It becomes linear. If this is not familiar, we will do this many times this semester,
so you need to chat with the TAs. TAs, can you raise your hands again? You got Logan,
Emily, Christian. If this is not at all familiar, the TAs will make sure this week to go over
how to read log plots and make them look linear because this will come back over and over
with diffusion, with creep, with lots and lots of things we talked about with plots,
so you need to learn how to do it. And it's not that bad. Just go meet with them. They'll
refresh you. Okay, so now that you got it linear, what you can do, let's rearrange our
equation, right? This was our same equation as before. We had dA dn equals A times the
range and our stress intensity raised to the n. If you bring the n over and you bring all
of this over here, this is called separating your variables, what it's going to allow us
to do is it's going to allow us to take the integral of both sides, right? We can take
the integral of the left-hand side and we can take the integral of the right-hand side.
The left-hand side has an integral of n, dn, right? The right-hand side, the integral is
dA. So we can take two different integrals. On the left-hand side, we're going to integrate
from the initial crack size, sorry, the initial, sorry, no cycles, right? Excuse me, from
the initial when you start the testing to the number of cycles to when it fails. And
that's really easy. This integral is easy to solve. It just equals nf. So the number
of cycles until it fails is going to be equal to this integral over here starting from your
initial crack size, A0. Remember, that's your half crack length too. That's not your
full crack. It's your half crack length. And it's going to increase until it reaches
the critical flaw size, right? When it goes from whatever it started at to the critical
flaw size, it will break, right? And so you can figure out this is pretty great because
we know how to calculate when things will break, remember? Because we said k1c, a fundamental
materials property, equals your y times your applied stress multiplied by pi A. We can
solve for the critical crack length, right? By just plugging in, let's say you know what
your stress is, your maximum stress. So if it's cycling between 500 and 200, it's going
to break when it's at the top of its cycle at 500, right? And then all you have to do
is solve for AC. So you know how big the biggest flaw can be before this thing fails. Is this
making sense, how this is all coming together? So now you just have to do this integral.
It's not that horrible because most of the things can come out of the integral. Only
things need to stay in the integral that have a dependence on A. So y technically is y as
a function of A over w, right? Remember that from before, that y we can use 1.12, but realistically
if A gets big, it would actually have to stay in the integral. So if I tell you that you
can assume y stays the same, go ahead and take it out. It'll make your life easier.
If on like the homework, you have to leave it in on one of the questions, then you have
to leave y in and you're going to have to do a much harder integral. Don't do that by
hand. Plug it into Wolfram Alpha or MATLAB or the software of your choosing, right? Just
do it by hand and print it off and put it on your homework, okay? Can I explain anything
up to now? Yeah. Great question. What do we use the whole make that power equation look
linear for? What you do that for is to figure out what big A and little n are, right? Because
you could figure out the slope of this line in this region 2, which is where the crack
growth equation works. Since you know the slope of that line, you can figure out n,
right? n is the slope. Once you know that, you can figure out capital A. So the whole
reason of plotting it this way is it allows you to figure out what your crack growth constants
are if I don't tell you them. I might just give you this figure. In fact, I think I do
on the homework, on one of the first homework questions. I've just given you a bunch of
data. You need to, from that data, figure out what capital A and little n are, your
crack growth constants. Once you know those, you plug in capital A and little n and all
these different values to calculate the number of cycles to failure. Are there any other
questions I can answer about this? We will do an example in just a moment. Yeah.
The scaling. Can you tell me what you mean? Oh, so this is plotted on a log scale, right?
These are not a linear scale. It's a log scale. Because of that, it allows us to plot this
whole term as just y. And because this is on a log scale, we can take this whole natural
log of delta k and just call it x. So it makes something that has an exponential, and it
makes it look linear. Yeah, aren't they? So there's lots of ways to solve for it. I
think the easiest is to figure out the slope. Some people will just take two points. They'll
take that point, and they'll take that point. And they say, I've got two equations. I've
got two unknowns. I can just bust this thing out. If you want to do that, that's great.
That works as well. Let's do an example. Here's a table of common crack growth constants
for a bunch of materials. So different materials, different metals. These basically are telling
you how your crack is going to grow in both metric and imperial, right? We're going to
use metric. We're not barbarians. OK, the question is this. We've only got six minutes.
We'll see how far we can get. A steel plate has a fracture toughness of 80 megapascal
root meters. It is alternately loaded between 500 megapascals in tension and then 60 megapascals
in compression. So it goes positive 50 to negative 60. If a and n are given, what is
the largest tolerable initial flaw size that it can be if your part has to survive 10 years
being cycled once every five minutes? Where do you even begin on a question like this?
Let's just work it together. Well, first off, let's figure out how many cycles it has to
survive. 10 years, right? It has to survive 10 years being once every five minutes. So how
many minutes are in 10 years? Well, in one year, there are 365 days. In one day, there
are 24 hours. In one hour, there are 60 minutes. And then we're going to do every five minutes
we have a cycle. OK, everything cancels out and you're left with just number of cycles.
So when I do this, I get 1.605 times 10 to the sixth cycles, right? So basically, you're
asking it to survive a million cycles, OK? OK, in order for it to survive a million cycles,
we've got that part of the question down. What else is it asking? It says, what is the
largest tolerable initial surface flaw that you can have since you know the crack growth
constant? Basically, that crack is going to grow. And after a million cycles, it better
be exactly at the critical flaw size, right? Or smaller. Make sense? So I'm asking you
to solve for A0 in this question. So let's go ahead and start plugging in things we know.
NF is 1.065 times 10 to the sixth cycles. That's going to be equal to 1 over A. Oh,
I don't remember. Let me pull it up. I did this before class. Oh, sorry, it's 1.05. 1.05.
It is to the positive sixth power, though. Sorry, 1.05. So it's a million cycles, OK?
We plug in A, right? A is equal to, in this case, 1.62 times 10 to the negative 12 meters
per cycle. So let's do 1.62 times 10 to the negative 12. That's going to be multiplied
by pi to the n over 2. n is 3.2. Divide that by 2, and it's 1.6.
The change in stress, what are we going to use for our delta sigma? What will we use
for our delta sigma? What's the range in stress? It's going from positive 500 down to negative
60, but the crack only grows during the positive portion. Yeah, question, Spencer?
Oh, I see.
Yeah, 0 to 500. So our range is just 500 megapascals. So 500 megapascals, that's raised
to the 3.2, right? We're going to take our integral from our initial flaw size up to
our critical flaw size, dA divided by y to the 3.2 multiplied by A to the 1.6. That's
3.2 divided by 2.
So we could solve this, except we need something else.
What do we need to know still?
Well, you have one equation so far, and we have two unknowns. We have A0, which we really
want to solve for, and we have AC. How do we figure out what AC is? Yeah, Rosa, question?
Exactly. You're going to use the fracture toughness equation. You're exactly right.
You're going to say that k1c equals y times sigma multiplied by the square root of AC.
We can solve for the critical flaw size by plugging in our maximum stress, which is going
to, when it fails, it's going to fail at 500 megapascals, right? It's going between negative
and up to 500. It's going to fail at the top of that cycle. We can plug in 1.12. We can
plug in the fracture toughness, which I think was 80. Yeah, 80 megapascal meters. So you
can solve for the critical flaw size in meters. When you do that, I end up with my critical
flaw size equals, I get 0.00649 meters. And we will finish this next time because we're
out of time. So this will be the first clicker question of the next class.