Latest / Mechanical Engineering Made Simple / Why Metals Break and How Engineers Fight Back

Transcript

Why Metals Break and How Engineers Fight Back · Mechanical Engineering Made Simple

0:00
Welcome back design junkies, to Mechanical Engineering Made
0:02
Simple. If the podcast brings you value,
0:05
help support the show with a like a follow or leave a comment
0:07
below. So imagine you're sitting in a
0:11
window seat on a commercial airliner, right?
0:14
You're cruising at, say, 35,000 feet.
0:17
Right way up there. Yeah, and outside it's like
0:19
-60°F, but inside, you know, you're enjoying a lukewarm
0:24
coffee. Totally comfortable in this
0:26
pressurized cabin. Best place to be?
0:28
Exactly. But I mean, right between you
0:30
and that freezing, totally oxygen deprived void is a sheet
0:35
of aluminum alloy that's barely thicker than a few credit cards
0:40
stacked together. It's actually terrifying when
0:42
you really think about it. It is because with every single
0:44
flight, that cabin is pumped full of air like a giant
0:47
metallic balloon which stretches the fuselage, and then it's
0:50
depressurized upon landing, relaxing so over years of
0:53
service, that thin skin expands and contracts 10s of thousands
0:57
of times. 10s of thousands. Right.
0:59
So how does the plane know not to just tear apart?
1:03
Like, how do the engineers who built it actually guarantee that
1:07
microscopic defects in the metal won't just suddenly link up and,
1:10
you know, unzip the aircraft at altitude?
1:12
Well, that's the multi $1,000,000 engineering question,
1:15
isn't? It it really is.
1:16
So today we are taking an absolute master class in the
1:20
unseen forces that dictate our physical world.
1:23
Our mission for this deep dive is to take this massive stack of
1:27
highly technical source material on material failure, fracture
1:32
mechanics, and fatigue and just, you know, break it down into the
1:35
ultimate real world lesson. I love it.
1:37
It's the perfect way to frame it because engineering isn't just
1:40
about building things, you know? It's fundamentally about
1:42
predicting how and when things will break.
1:45
Right, predicting the failure. Exactly.
1:46
Whether you are designing those aircraft engine components,
1:50
pouring concrete for suspension bridges, laying deep sea
1:53
pipelines, or I mean, if you're just insanely curious about why
1:56
the glass screen on your phone shatters into a spider web while
1:59
a paper clip can bend back and forth a dozen times before
2:02
snapping. Yeah, the paper clip thing is
2:03
clap. Right, but the underlying
2:06
physics are identical O We are going to dig into the microscoic
2:10
battlefield inside these materials to see exactly how
2:14
structural failure happens. Consider this your orientation,
2:17
everyone. We have a massive amount of
2:19
ground to cover. We're going to explore exactly
2:21
what dictates whether a material stretches like warm putty or
2:24
violently snaps like a dry twig. Which is a crucial distinction.
2:27
Huge. And we'll break down the silent,
2:30
insidious killer of engineering structures known as cyclic
2:33
loading. Then we will explore how
2:36
engineers actually fight back by intentionally pre stressing
2:39
materials to basically cheat physics.
2:42
Which is always fun. Oh yeah, and finally, we'll zoom
2:45
all the way down to the atomic lattice itself to see how the
2:48
literal size of the metal grains changes the physical path of a
2:51
crack. And we are going to do it
2:52
without relying on the dense, unreadable jargon that usually
2:55
surrounds this field. You know, we want you to
2:57
understand the why and the how, not just memorize formulas from
3:00
a textbook. Exactly.
3:02
So I want to kick things off with a very relatable
3:05
observation. Let's say you go into a hardware
3:07
store, right, And you buy a thick piece of plain carbon
3:11
steel wire. OK, got it.
3:13
You take it home, grab it with two pliers and you apply
3:16
pressure. It slowly bends.
3:19
It yields to the force. Right, it deforms.
3:21
Yeah, but now you take a stick of standard classroom chalk, you
3:26
apply that exact same bending pressure, and it doesn't bend at
3:29
all. There is zero warning.
3:30
It just boom violently snaps in half.
3:32
Right, a totally different failure mode.
3:34
Yeah. So what is the fundamental, like
3:36
atomic difference between those two materials that causes such
3:39
drastically different behavior? Well, it all comes down to their
3:43
internal atomic architecture. Specifically something we call
3:46
dislocation mobility. Dislocation mobility.
3:48
OK. Let's look at the chalk first.
3:50
Chalk is what we call an ionic solid.
3:52
The atoms inside it are locked into this really rigid
3:55
alternating grid of positive and negative charges.
3:58
Like a lattice structure. Exactly.
4:00
And because of that rigid ionic bonding, there is virtually 0
4:04
dislocation mobility. OK, wait, let's make sure we
4:07
have a really clear picture of a dislocation for everyone.
4:10
That's essentially a tiny structural defect, right?
4:12
Like a missing line of atoms in the crystal lattice.
4:15
Precisely. I mean, no crystal is perfect in
4:18
ductile materials like metals. You have these dislocations, and
4:21
critically, those dislocations can move.
4:24
No, they actually shift around. Yes, when you apply stress to a
4:28
piece of steel, entire planes of atoms can actually slide past
4:32
one another along these defects. Wow, OK.
4:35
And this sliding action is what allows the metal to permanently
4:38
bend without breaking. It's the materials way of
4:41
gracefully relieving the stress you're putting on it.
4:44
But the chalk can't do that. Right, in an ionic solid like
4:48
chalk, the atoms are locked down.
4:50
The internal structure simply cannot slide.
4:52
Oh, it's kind of like a massive traffic jam on a highway that's
4:55
surrounded by concrete barriers on both sides.
4:58
Well, that's a great way to think about it.
4:59
Right, because if the cars are packed bumper to bumper and
5:02
there is absolutely no shoulder to merge into, you have 0
5:05
mobility. You can't relieve the
5:06
congestion, so if a truck comes barreling into the back of the
5:10
line, the force has nowhere to go.
5:11
You just get an immediate catastrophic crash.
5:14
That is a brilliant analogy, yeah.
5:16
And because the chalk has no shoulder to pull into, because
5:19
it literally cannot deform to absorb the energy, the primary
5:23
forces driving the crack are the stresses acting directly
5:27
perpendicular or, you know, normal to the crack face.
5:30
OK, so pushing straight against it, yeah.
5:32
So if you take a standard cylinder of chalk and you twist
5:35
it applying torsion, the principal stress is pulling the
5:38
material apart are oriented at exactly a 45° angle to the axis
5:43
of the chalk. Oh wow wait is that why when you
5:45
twist a piece of chalk it always breaks with that weird angled
5:48
spiral looking surface? Exactly that helical fracture
5:51
surface is the pure physical signature of a brittle ionic
5:56
solid failing under tension. That is so cool, I've always
5:59
wondered why it snaps like that. Yeah, now contrast that chalk
6:03
with the material on the complete opposite end of the
6:05
spectrum. Think of something like
6:07
plasticine or. Modeling clay, right?
6:09
If I pull on modeling clay, it doesn't snap.
6:11
It just stretches and thins out until it finally pulls apart
6:15
into two soft cones. Exactly because plasticine is a
6:19
polymer capable of massive plastic deformation, its
6:23
internal structure is completely different from the chalk's rigid
6:26
grid. It allows molecules to slide
6:29
past one another incredibly easily.
6:31
So it's basically all shoulder, no traffic.
6:33
Yeah, basically. So its failure is entirely
6:36
ductile and occurs via shear. The material literally slides
6:40
apart. Oh, interesting.
6:41
Yeah, and in this case the maximum shear stresses act
6:45
perfectly perpendicular to the axis of the specimen.
6:47
So you get a flat sliding failure profile, which is
6:50
completely different from the angled snap of the jock.
6:53
OK, that brings us to a massive question that engineers have to
6:56
grapple with every single day. Obviously lain carbon steel
6:59
isn't modeling clay, but it isn't shock either, right?
7:02
Right, it sits somewhere in the middle.
7:04
Exactly. So how do we control this?
7:06
Like how do we dictate when an industrial material like the
7:09
steel used in a bridge axed ductile versus when it acts
7:13
brittle? Well, we define this behavior
7:16
using a concept called the Ductile brittle transition
7:18
Temperature or DBPT. DBTT OK.
7:21
Right. Materials fundamentally change
7:23
their personality based on temperature.
7:26
You might have a steel that is wonderfully tough and ductile at
7:28
room temperature, but you know if you drop it to freezing it
7:32
suddenly acts like glass. Which is terrifying for winter
7:34
construction. Oh.
7:35
Absolutely, but the exact temperature where this
7:38
transition happens isn't fixed. It is heavily influenced by
7:42
three main physical pillars, plus the actual physical
7:45
geometry of the part itself. OK, let's break down those
7:48
pillars. I want to understand this.
7:49
Let's start with pillar #1, the chemical environment.
7:52
This is a huge one, especially for like the oil and gas
7:55
industry. Introducing what we call
7:58
embrittling agents into the metals environment artificially
8:01
raises the transition temperature.
8:03
So it makes it brittle at higher temperatures than normal.
8:05
Exactly. And that the most notorious and
8:08
frankly terrifying of these agents is hydrogen.
8:11
Wait, hydrogen like the most abundant element in the
8:14
universe? Why is that so dangerous to
8:16
steal? Because hydrogen atoms are
8:18
incredibly tiny. I mean, they're basically just
8:21
single protons. So when a steel pipeline is
8:24
exposed to certain corrosive environments or even, you know,
8:26
during the welding process, these tiny hydrogen atoms can
8:30
easily diffuse right into the solid metal lattice of the steel
8:34
they. Just slip right in, yeah?
8:35
They sneak right between the iron atoms, and once inside,
8:39
they tend to congregate at the tips of microscopic flaws or
8:42
cracks. So they basically gang up inside
8:45
the metal. What do they actually do once
8:47
they're in there? Hiding the cracks.
8:48
They interfere with the atomic bonds.
8:50
It's called decohesion theory. The hydrogen lowers the cohesive
8:54
energy holding the iron atoms together, and simultaneously it
8:58
restricts that crucial dislocation mobility we talked
9:01
about earlier. It jams up the traffic again.
9:03
Exactly. It jams up the traffic.
9:05
So a steel pipeline carrying hydrogen or one exposed to the
9:08
deep ocean will behave as if it's much colder and much more
9:12
brittle than it actually is. Yeah, it might fail
9:15
catastrophically under a normal operating pressure simply
9:18
because it's been poisoned by hydrogen.
9:19
That is wild. You can literally poison a metal
9:23
to make it brittle. OK, let's move to pillar #2 the
9:25
strain rate. Now, in layman's terms, this is
9:28
basically how fast you are hitting the material, right?
9:31
That's a great way to put it, yeah.
9:33
Increasing the strain rate, meaning you apply the physical
9:36
load, very suddenly increases the transition temperature,
9:40
pushing the material toward brittle behavior.
9:42
OK, so faster impact equals more brittle?
9:44
Exactly. If we connect this back to our
9:46
traffic jam analogy, it makes perfect sense.
9:49
Ductile behavior requires those dislocations, those crystal
9:52
defects, to physically slide and move through the metal to
9:55
relieve the stress. Right, the cars need time to
9:58
merge into the shoulder. Yes, that atomic sliding is a
10:01
physical process and it requires time to happen.
10:04
You know, it makes me think of swimming, or well more
10:07
specifically, slapping the water well.
10:09
I like where this is going. Like if I take my open hand and
10:12
slowly push it into a pool, the water molecules easily yield and
10:16
flow around my hand, right? Feel soft.
10:18
Right, because you're moving slowly, yeah.
10:20
But if I raise my hand and slap the surface of the pool as fast
10:24
and as hard as I can, it feels like hitting a sheet of
10:27
concrete. My hand actually stings because
10:30
I didn't give the water molecules time to move out of
10:33
the way. That is a phenomenal analogy.
10:35
It is the exact same principle in SOLID mechanics.
10:38
If you apply a slow heavy load to a steel beam, the atomic
10:42
planes have time to slip the material bends.
10:45
But if it's sudden. Right.
10:47
If you hit that exact same steel beam with a sudden massive
10:50
explosive impact, the dislocation simply do not have
10:53
sufficient time to mobilize in response to the load.
10:56
The material has absolutely no choice but to snap.
10:59
It's all about time. OK, Which brings us to pillar #3
11:03
alloy composition. What are we actually mixing into
11:06
the steel in the Founder? Let's stick with our plain
11:08
carbon steel example. Increasing the concentration of
11:11
carbon in the steel increases the transition temperature.
11:15
It makes it more brittle. OK, why does adding carbon do
11:17
that? The reason is purely mechanical.
11:21
Carbon atoms are what we call interstitial, meaning they don't
11:24
replace iron atoms in the grid. They sit in the empty spaces,
11:28
the gaps between the larger iron atoms.
11:30
Oh, I see. So they're like like boulders
11:33
sitting in the middle of our highway shoulder.
11:35
Exactly. When you add more carbon, you
11:38
are physically placing roadblocks in the path of the
11:41
dislocations. The carbon physically pins the
11:44
dislocations down, inhibiting the motion required for ductile
11:48
deformation. So the traffic can't merge at
11:51
all, right? And the trade off here is
11:52
legendary. In engineering, adding carbon
11:55
gives you a much stronger, harder steel that resists
11:58
bending, but one that is infinitely more likely to
12:01
shatter into pieces when it finally is pushed past its
12:04
limit. It's like a blacksmith making a
12:06
sword. You want it hard so it stays
12:08
sharp, but not so hard that it shatters when you hit a shield.
12:11
Exactly. It's a delicate balance.
12:13
Now, you also mentioned earlier that geometry plays a role.
12:16
This one confuses me a bit. Why would a thick plate of steel
12:20
behave differently than a thin sheet if they are poured from
12:23
the exact same batch of liquid metal?
12:25
It is entirely counterintuitive, I'll give you that.
12:28
Until you visualize the stresses.
12:31
It's the difference between what structural engineers call plain
12:34
stress and plain strain. OK, break that down for me.
12:38
Sure. Thick, heavy plates operate
12:41
under plain strain conditions, and they are significantly more
12:44
prone to brittle fracture. Wait, why wouldn't thicker be
12:47
stronger and, you know, safer? Stronger, Yes.
12:50
Safer against brittle fracture? Absolutely.
12:53
No. Really.
12:54
Yeah. Imagine a tiny crack forming
12:56
deep inside a very thick plate of steel.
12:59
When the metal right at the tip of that crack tries to stretch
13:01
and yield to relieve the stress. It can't.
13:04
Because there's too much metal around it.
13:06
Exactly. It is completely surrounded,
13:07
boxed in by massive amounts of other solid metal.
13:11
It can't easily shrink or pull inward in the thickness
13:14
direction. We call this a triaxial
13:15
constraint. Triaxial constraint.
13:17
Got it. And because it is physically
13:19
constrained from yielding, it behaves like our chalk.
13:23
It snaps. Oh, but a thin plate has an
13:25
escape route. Exactly.
13:27
For thin plates, you are in a state of plane stress.
13:30
The size of the deformation zone at the tip of the crack is
13:34
roughly equal to the thickness of the plate itself.
13:37
Because the outer surface of the plate is right there, just, you
13:40
know, millimeters away. The material has freedom.
13:43
It can yield, pull inward and physically thin.
13:46
I like striking a piece of taffy to relieve the stress.
13:49
Oh, that makes total sense visually.
13:51
Yeah. Therefore, a failure in a thin
13:53
plate is almost always going to occur by slow ductile tearing,
13:57
giving you plenty of warning rather than a sudden,
14:00
catastrophic brittle snap. So when we step back and look at
14:03
this whole ductile, brittle thing, ductility isn't just a
14:06
magical inherent property printed on a spec sheet.
14:09
None at all. It's a constant, active
14:11
negotiation between the environment the part lives in,
14:14
how fast the world impacts it, the chemical roadblocks we've
14:18
baked into its atomic lattice, and like, how thick we actually
14:22
decided to cast it. Which is exactly why
14:24
catastrophic failures still happen.
14:26
I mean, an engineer might design A crane hook perfectly for a
14:28
static, slow, heavy load in a warm, sunny shipyard, and it
14:33
behaves beautifully. But take that exact same crane
14:36
to a freezing offshore oil rig in the North Sea, expose it to a
14:41
corrosive salt environment, and subject it to sudden swinging
14:45
impacts from ocean waves. That perfectly designed hook
14:48
will shatter like dropped glass. That sets the stage perfectly.
14:52
I mean, everything we've discussed so far revolves around
14:55
a single heavy load, like a one time extreme event, correct?
14:59
But what happens when we aren't dealing with a massive impact,
15:01
but rather a tiny, seemingly harmless load applied and
15:05
removed over and over and over again that takes us into the
15:09
realm of the hidden killer of engineering fatigue?
15:12
Fatigue is endlessly fascinating to me because it completely
15:15
defies human intuition. How so?
15:17
Well, our brains are wired to think if it can hold 100 lbs
15:19
once, it can hold 100 lbs forever.
15:21
Yeah, that seems logical. Right, but a steel component
15:24
might easily and comfortably survive a static resting load.
15:29
You hang away from it and it holds indefinitely.
15:31
But you take that exact same weight, a load that is
15:34
mathematically well below the materials yield strength, and
15:38
you apply it, then remove it, apply it, remove it thousands of
15:42
times. It breaks.
15:43
It can and will fail catastrophically, often with 0
15:47
visible warning. I think literally everyone
15:48
listening has experienced this in real life when trying to
15:52
break a paper clip or, you know, a zip tie or a piece of wire.
15:57
Oh, the paper clip is the classic fatigue test.
15:59
Right, you grab it, you pull it with all your might, and you
16:01
can't break it with your bare hands.
16:03
It's just too strong. So what do you do?
16:04
You bend it back and forth, up and down, maybe 50 times.
16:08
Sometimes it gets hot and then suddenly it snaps with almost no
16:11
effort. That is the absolute essence of
16:13
cyclic loading, and engineers recognize this phenomena long
16:17
before we had modern scanning electron microscopes or advanced
16:21
fracture mechanics. Really, how far back does this
16:23
go? Back in the 1840s and 50s with
16:26
the explosion the railway industry solid steel train axles
16:30
were suddenly snapping in half mid journey.
16:33
Man, that's bad. Yeah, it was disastrous.
16:35
And these axles were massively overbuilt.
16:37
By all static calculations, they should have lasted forever.
16:41
But a German engineer named August Voller was one of the
16:44
first to rigorously investigate this.
16:46
He developed empirical testing machines, literally rotating
16:50
bending tests to figure out why these axles were failing.
16:53
So he essentially built a machine to bend giant metal
16:58
paper clips until they broke just to track the data.
17:03
What we did, and to understand what Voler found, we need to map
17:06
out the vocabulary of cyclic loading.
17:08
We have to define the parameters that characterize A fatigue
17:11
history. OK, let's go.
17:12
Imagine looking at a graph on an oscilloscope.
17:14
The vertical axis is stress, how hard we are pulling or pushing,
17:18
and the horizontal axis is time. The stress goes up and down in a
17:22
continuous wave. OK, I'm visualizing the wave.
17:24
What are the specific measurements engineers actually
17:26
care about here? The 1st and most obvious is the
17:29
stress range. This is simply the absolute
17:33
maximum stress at the peak of the wave minus the minimum
17:36
stress at the bottom of the through.
17:37
It is the total physical height of the wave.
17:39
Makes sense. From the absolute lowest valley
17:42
to the highest mountain peak. What's next?
17:44
Next is the stress amplitude. This is exactly 1/2 of the
17:48
stress range. Cast the range right.
17:50
So if your wave fluctuates perfectly evenly around zero,
17:54
say you're pulling with 50 lbs of force, then pushing with 50
17:59
lbs of force, the stress amplitude is simply 50 is how
18:02
far the wave deviates from its center line.
18:05
Got it. And then we need to know where
18:06
that center line actually sits, right?
18:08
Which I assume is the mean stress.
18:10
Correct. Mean stress is 1/2 of the sum of
18:13
the maximum and minimum stress. It's the baseline that the wave
18:16
oscillates around. OK, give me an example of that.
18:18
If you pull on an airplane wing with 100 units of force, and
18:21
then you only relax it down to 50 units of force, the wave is
18:25
riding entirely in positive territory.
18:27
The mean stress in that case is 75.
18:29
It never goes into compression. OK, I follow and finally you
18:32
have the load ratio. Right, known in equations as R,
18:36
the load ratio is a quick shorthand.
18:38
It is simply the minimum stress divided by the maximum stress.
18:41
So it's just a quick identifier for engineers.
18:44
Exactly. Yeah, it instantly tells another
18:46
engineer what kind of fatigue cycle you're dealing with.
18:49
If someone tells you the load ratio is -1 you immediately know
18:54
the minimum stress is the exact mathematical negative of the
18:57
maximum stress. It is a fully reversed,
19:00
perfectly balanced cycle. No, wait, I have to push back.
19:03
Here we've defined the height of the wave, the middle of the
19:06
wave, the bottom of the wave. But what about the shape of the
19:09
wave or the speed? Good question.
19:12
Like, does the frequency matter if I bend that paper clip back
19:15
and forth 10 times a second versus 100 times a second?
19:17
Or what if it's a smooth, gentle sine wave versus a really harsh,
19:22
jerky square wave? Doesn't a faster, harsher
19:25
vibration cause more damage? It is an incredibly logical
19:28
question, and the answer almost always shocks people when they
19:31
first learn it. For most structural metals
19:33
operating at room temperature in a normal dry atmosphere,
19:37
frequency and waveform generally do not influence fatigue life.
19:40
Wait, really? A cycle is a cycle whether it
19:43
takes one second or one minute. Purely mechanically, yes.
19:46
The microscopic damage to the atomic lattice is done by the
19:50
physical reversal of the strain. The actual sliding of those
19:54
atomic planes back and forth, the speed at which they slide,
19:57
doesn't significantly change the amount of damage per cycle.
20:00
I feel like there has to be a catch.
20:02
There is a massive caveat. Frequency and waveform become
20:05
incredibly important, often dominating the failure if
20:09
environmental effects are present.
20:11
The real world ruins the math again.
20:13
Exactly. If you are operating a pump in a
20:16
highly corrosive chemical bath, or in high humidity or at
20:19
elevated temperatures like inside a jet turbine where the
20:22
metal might slowly creep and deform over time, then the clock
20:25
matters immensely. Why does the clock matter then?
20:28
Because a slower frequency means the crack tip is exposed to that
20:32
corrosive acid or that extreme heat for a longer duration per
20:35
cycle, the environment has more time to attack the fresh metal
20:39
exposed at the crack tip. Oh, that makes perfect sense.
20:42
Right. But if we are testing a steel
20:44
rod in a perfect vacuum at room temperature, speed doesn't
20:48
matter. The cycle cap is all that
20:50
matters. So how do engineers actually use
20:53
this wave data to predict when a bridge cable or an axle is going
20:57
to snap? Like how do they predict the
20:58
future? They map it out using what is
21:00
famously known as an SN curve. SN Curve, yeah.
21:04
S stands for the applied stress amplitude and N stands for the
21:08
number of cycles to failure. You plot stress on the vertical
21:11
axis and cycles on the horizontal axis.
21:13
OK, pretty straightforward. And because a part might survive
21:16
10 million cycles, the horizontal axis is logarithmic
21:20
just so we can fit it on a single piece of paper.
21:22
So the testing process is basically take a pristine metal
21:25
rod, hit it with a massive stress amplitude, It breaks in
21:29
just 1000 cycles. You log that dot on the graph,
21:32
Take a new rod, hit it with a slightly lower stress, it
21:35
survives 50,000 cycles. You plot that dot.
21:39
You keep doing this and you map out this curve sloping downward.
21:43
The lower the stress, the longer it lives.
21:45
Exactly. You're building a historical
21:46
map. But here is the HA moment of
21:50
this whole field. The curve doesn't just go down
21:53
forever, does it? It does not, and this phenomenon
21:56
is an absolute lifesaver for civilization.
21:59
For many materials, specifically most plain carbon steels and
22:03
titanium alloys, as you drop the stress lower and lower, the SN
22:07
curve eventually flattens out entirely.
22:09
It becomes a perfectly horizontal line and.
22:11
What do we call that? This threshold is called the
22:13
endurance limit or the fatigue limit.
22:15
And what does a perfectly flat horizontal line actually mean
22:18
for the physical metal? It means that if you keep the
22:21
applied cyclic stress strictly below that specific endurance
22:24
limit, the component has an effectively infinite life.
22:27
Infinite life, literally forever.
22:29
From a fatigue standpoint, yes. The micro damage that is created
22:33
during the loading cycle is so miniscule that it is entirely
22:37
arrested by the materials, internal microscopic barriers,
22:41
the grain boundaries. You just stops the cracks in
22:43
their tracks. Right, the damage simply cannot
22:46
accumulate into a fatal crack. The material heals itself in a
22:50
way on every single cycle. That is incredible.
22:54
Is there a rule of thumb for where that magical infinite life
22:57
threshold sits? Like how much do I have to D
23:00
rate the metal to be safe? Generally for steels, this
23:03
endurance limit hovers around 35 to 50% of the materials ultimate
23:07
tensile strength. OK.
23:08
So about half. Roughly yes.
23:10
So if your steel alloy snaps under a static pole of 1000
23:13
megapascals, its endurance limit for cyclic vibration will be
23:16
somewhere around 400 to 500 megapascals.
23:19
If you keep the vibrations below that, it runs forever.
23:21
But wait, do all materials have this infinite life threshold?
23:25
Because if I'm building an airplane out of aluminum, does
23:27
aluminum have an endurance limit?
23:28
Unfortunately no, and this is a critical distinction in
23:31
aerospace. Many non ferrous alloys,
23:35
specifically aluminum and copper, do not exhibit A
23:38
distinct endurance limit. Wait, they don't.
23:40
No, their SN curve just keeps sloping downward very slowly
23:45
forever, no matter how tiny the vibration.
23:48
If you cycle an aluminum wing enough times, it will eventually
23:51
crack. So how do you design an airplane
23:53
if you know the wings are guaranteed to eventually fall
23:55
off? That sounds terrifying.
23:57
It sounds scary, but it's handled very carefully.
24:00
For materials without a true endurance limit, engineers
24:03
arbitrarily define one. By convention, they establish A
24:06
fatigue strength based on a massive number of cycles,
24:09
usually 10 million or sometimes 500 million cycles for aerospace
24:12
parts. If the aluminum part can survive
24:15
a lifetime of flakes, that equates to 10 million stress
24:17
cycles. We deem it safe and we Simply
24:20
put a mandatory retirement age on the aircraft frame.
24:23
OK, that makes me feel slightly better about flying.
24:25
So we have these curves mapped out, but engineers can't just
24:28
stare at graphs all day. They need software, they need
24:31
math. They need equations to calculate
24:34
exactly when the part will fail when it's operating on the
24:37
sloped part of the SN curve. For that, we rely on the stress
24:40
life approach, which is governed by the Baskin relationship
24:44
proposed by OH Baskin back in 1910.
24:46
He looked at Villers data and realized that if you plot the
24:50
logarithm of the stress amplitude against the logarithm
24:53
of the cycles to failure, that curved line straightens out into
24:57
a perfect diagonal line. Let's translate this math into
25:00
concept, because reading raw equations on a deep dive is a
25:03
great way to put people to sleep.
25:05
How does this relationship actually work?
25:07
What is the spoken math here? Let's keep it conversational.
25:10
Baskin's relationship basically maps out of balance scale for
25:13
survival. The equation is basically delta
25:16
Sigma divided by 2:00 which gives us the stress amplitude
25:20
and that equals the materials fatigue strength coefficient
25:23
multiplied by the quantity 2 times NF, all raised to the
25:26
power of B. OK, so on one side of the Ledger
25:28
you have your stress amplitude, how hard you are bending or
25:31
pulling the metal. Right.
25:33
And on the other side you have the materials innate fatigue
25:37
strength coefficient, which is basically its true strength in a
25:41
single massive pull scaled by the number of times you're
25:44
cycling it, which is the NF part.
25:45
And the crucial part is that the number of cycles is tied to an
25:48
exponent. Right that back you mentioned
25:50
the Baskin exponent. Exactly because it's an
25:53
exponential relationship governed by that Baskin
25:55
exponent, it means fatigue is incredibly sensitive to the
25:58
load. How?
25:59
Sensitive. Extremely a tiny 10% increase in
26:04
the stress you are applying doesn't just reduce the parts
26:06
life by 10%, it might slash the number of cycles it can survive
26:10
by half or more. It drops off a Cliff.
26:13
So if I know my metals baseline strength, and I know it's
26:16
specific exponent which tells me how steep that Cliff is, I can
26:20
plug in my operating vibration and know exactly how many days
26:23
or hours the part will survive. That is the foundation of high
26:26
cycle fatigue design. But as I'm about to point out,
26:29
this perfectly mathematically predictable world only exists in
26:33
a laboratory where the loading is perfectly balanced around 0.
26:37
Oh, absolutely. Exactly.
26:38
The laboratory tests assume you pull with 10 lbs, then push with
26:42
10 lbs. Mean stress is 0 but real life
26:46
is messy. What if the suspension bridge
26:48
cable is already pre tension to hold up thousands of tons of
26:51
concrete and then the wind starts vibrating it?
26:54
The cable is constantly under tension.
26:56
The resting state isn't 0. Right and pulling a material
26:59
tight before you start vibrating it drastically accelerates
27:03
fatigue failure. The mean stress is non zero and
27:06
usually positive. To account for this, engineers
27:09
over the decades develop mean stress correction models.
27:12
How do they work? Do they just artificially lower
27:14
the endurance limit? Essentially, yes.
27:16
These models take the safe stress amplitude you calculated
27:19
for a perfectly balanced test and they penalize it.
27:22
They shrink your safe operating window based on how much static
27:25
mean stress is constantly sitting on the part.
27:28
There are three famous historical models used for this
27:30
Soderbergh, Goodman and Gerber. Let's walk through them.
27:34
Let's start with Soderbergh. The Soderbergh model was
27:36
developed for the paranoid engineer.
27:38
It is incredibly conservative. Paranoid engineer.
27:41
I like that. It essentially says that as your
27:43
static mean stress climbs closer to the materials yield point,
27:48
the point where it starts to permanently bend, your allowable
27:51
safe cyclic vibration drops in a straight sheer line down to 0.
27:56
So it's super strict. Very.
27:59
If you are resting at the yield point, you cannot tolerate any
28:02
vibration at all. Better safe than sorry I guess.
28:05
But if I'm building an airplane, better safe than sorry.
28:08
Means the plane is so heavy with extra metal that it can't take
28:11
off. Exactly.
28:13
Soderbergh often results in parts that are unnecessarily
28:16
bulky, so the aerospace and automotive industries often turn
28:20
to the Goodman model. How does Goodman give us more
28:22
breathing room? Goodman is less conservative.
28:25
It uses a very similar linear penalty, but instead of drawing
28:29
the line down to the yield point, it draws a line all the
28:31
way out to the ultimate tensile strength, the point where the
28:34
metal actually rips into two pieces.
28:36
Oh, that's a big difference. Yeah, because the ultimate
28:38
tensile strength is much higher than the yield point, the
28:41
Goodman model gives the engineer a much larger envelope of
28:46
allowable vibrations to work within.
28:48
So it's a bit riskier, but it saves weight.
28:50
What about the third one, Gerber?
28:52
The Gerber model throws out the straight line entirely.
28:55
Experimental data, particularly for highly ductile alloys under
28:59
tension, show that the failure points didn't drop in a straight
29:02
line, right? They bulged outward in a curve.
29:05
So Gerber uses a parabola to map the penalty.
29:08
A curve gives you even more operating room in the middle
29:10
ranges. It does, and it's highly
29:12
accurate for tensile mean stresses, but there is a
29:15
mathematical flaw in the Gerber model.
29:17
What's the flaw? Because the formula uses the
29:19
square of the mean stress to create that parabola.
29:22
A negative mean stress, meaning the part is under constant
29:25
compression, becomes positive when you square it.
29:28
The mass penalizes you for compression.
29:30
But wait, earlier we said squeezing a crash shut is good.
29:33
Exactly, compressive mean stresses are incredibly
29:36
beneficial and actually extend fatigue life.
29:39
So the Gerber model is physically wrong for compressive
29:42
loads. You should never use it if your
29:44
part is being squeezed, only if it's being stretched.
29:46
OK, so we've solved for unbalanced loads.
29:48
We know how to penalize the math.
29:50
But here is the ultimate real world problem.
29:53
If I'm driving a heavy 18 Wheeler truck over a steel
29:56
bridge, sometimes the truck is fully loaded with cargo, maximum
30:01
stress wave. Other times the truck is totally
30:04
empty, minimum stress wave. Sometimes it's a calm day,
30:06
sometimes there are massive crosswinds.
30:08
The bridge is experiencing wildly different amplitudes of
30:11
stress every single minute of the day.
30:13
It's a totally chaotic history. It is chaos.
30:15
So how on earth do you add up the different amounts of damage
30:19
to know when the bridge will collapse?
30:20
This is perhaps the most critical challenge in structural
30:23
engineering quantifying cumulative damage.
30:26
The universal standard, despite its flaws, is the Pongren minor
30:30
linear damage rule. Pongren Minor.
30:32
I love the concept of this because it's so intuitively
30:35
simple. Walk us through how we calculate
30:37
the damage. It's brilliantly simple.
30:39
It relies on fractions. Imagine a steel component from
30:42
our SN curves. We know that if we subject it
30:45
strictly to a high stress, it's called Stress level A.
30:49
It will snap it exactly 10,000 cycles.
30:53
Now in the real world, the part only experiences 1000 cycles of
30:57
stress level A before the load changes.
31:00
What have you done? You have consumed exactly 110th
31:03
or 10% of its total life. Right, a simple fraction 1000 /
31:07
10,000. Correct.
31:09
Now the wind dies down, the load drops to stress level B.
31:13
At this lower stress, the SN curve says the part could
31:16
normally survive 100,000 cycles. Our real world part runs for
31:21
20,000 cycles at this lower level.
31:23
So 20,000 / 100,000, that's 20%. You've consumed 20% of its life
31:28
at the lower stress. Exactly.
31:29
Palmgren Miner simply tells you to add the fractions 10% from
31:32
the high stress plus 20% from the low stress.
31:35
You have consumed 30% of the components total fatigue life.
31:38
You keep monitoring the loads, adding up the fractions and.
31:40
When do you hit failure? When the sum of all those
31:43
fractions reaches 1 or 100%, the math says the component breaks.
31:49
It is elegant. It's perfectly linear.
31:51
But you said despite its flaws, what is the catch?
31:55
Does the order of the loads matter?
31:56
Like what if the truck is heavy on Monday and light on Tuesday
32:00
versus light on Monday and heavy on Tuesday?
32:02
That is the massive fatal flaw of the rule.
32:05
The Palmgren minor equation inherently assumed that the
32:08
sequence in which the loads are applied has absolutely zero
32:12
influence on the outcome. Which mathematically makes
32:14
sense. Addition is commutative.
32:16
Right, 10 + 20 is the same as 20 + 10, but in the physical
32:19
reality of metallurgy, the sequence of loads matters
32:22
immensely. Explore that.
32:24
Give me a physical example of how the sequence changes the
32:27
survival of the metal. Let's take case 1.
32:29
High stress first, low stress second.
32:32
You apply a massive, violent cyclic load to a pristine part.
32:36
Because the stress is so high, it easily initiates microscopic
32:40
cracks at the surface of the metal.
32:41
OK, damage is done. Then you lower the load.
32:44
Now you are applying A gentle low stress cycle.
32:46
But the cracks are already there.
32:48
Precisely that gentle low stress might never have been strong
32:52
enough to initiate a crack on its own, but because the high
32:56
stress already opened the door, the low stress is perfectly
32:59
capable of propagating those existing micro cracks.
33:03
I see. The crack physically
33:05
concentrates the stress right at its impossibly sharp tip, so the
33:09
damage accelerates rapidly. In this high low scenario, the
33:13
part will fail much earlier than Palmgren Minor predicts.
33:16
The sum of the fractions might only reach 0.7 before the part
33:20
snaps. So the math says you have 30%
33:23
life left, but the part is already broken on the floor.
33:26
That's terrifying. Now let's reverse it.
33:28
Case 2. Low stress first, high stress
33:32
second. In case 2, you apply the gentle
33:34
low stress cycle first. It isn't high enough to cause
33:37
any micro cracks, but what it does do is manipulate the
33:40
internal lattice. It causes dislocations to move
33:43
and tangle up around each other. We call this strain hardening.
33:46
Chain hardening, yes. The surface of the metal
33:49
actually becomes physically harder and stronger.
33:51
It's literally like an athlete warming up, conditioning their
33:53
muscles before the heavy lift. It's a perfect analogy.
33:57
The metal is conditioned, then when you finally apply the
34:00
massive high stress load, it is attacking a material that has
34:04
been hardened by the previous cycles.
34:06
So it puts up a better fight. It is now significantly more
34:09
difficult for that high stress to initiate surface damage.
34:13
In this low high scenario, the part actually lives much longer
34:17
than the math predicts. The Palmgren miner sum might
34:20
reach 1.5 before failure. So the history, the actual lived
34:24
experience of the metal fundamentally alters its future.
34:28
I love that. Now, Speaking of the lived
34:30
experience of the metal, everything we have discussed so
34:32
far assumes the metal is staying mostly elastic.
34:35
It's bringing back to its original shape like a rubber
34:38
band. Yes, mostly elastic.
34:39
But what happens when we push the material so violently that
34:43
it deforms permanently on every single cycle?
34:45
It doesn't spring back, it yields.
34:48
That brings us to a massive fork in the road for engineering
34:50
design. High cycle fatigue versus low
34:53
cycle fatigue. Let's define the boundary.
34:55
High cycle fatigue is what we've been talking about, correct?
34:58
The stress life approach, Bowler's curves, Baskin's
35:01
relationship, these are all built for high cycle fatigue, or
35:05
HCF. They apply to situations where
35:07
the stresses are relatively low, the deformation is almost
35:10
entirely elastic, and you expect the component to survive 10s of
35:14
thousands or millions of cycles. But sometimes a component is
35:18
subjected to absolute hell. I'm thinking of the turbine
35:21
blades inside a jet engine experiencing massive thermal
35:24
expansion, or the metal around a sharp bolt hole in an excavator
35:29
bucket. In those extreme situations,
35:31
significant plasticity is involved.
35:34
The material isn't just springing back, it is physically
35:37
yielding. It is stretching and compressing
35:39
permanently back and forth on every cycle.
35:41
That sounds rough on the metal. It is when this happens the
35:44
components lifetime is going to be incredibly short, maybe just
35:47
a few 100 or a few thousand cycles.
35:49
This is low cycle fatigue or LCF.
35:52
And I assume our beautiful stress life equations completely
35:54
breakdown here. They collapse entirely.
35:57
Stress is no longer a reliable indicator of the damage being
36:00
done. Why not?
36:01
Because once a metal yields and goes plastic, a tiny, almost
36:04
unmeasurable increase in stress causes a massive sweeping
36:08
increase in physical strain. You can't track the stress
36:11
accurately anymore. So if we can't measure the
36:14
invisible force, the stress, what do we measure?
36:17
We measure the physical reality. We use a strain life approach.
36:21
Rather than characterizing the fatigue load by stress
36:23
amplitude, we measure the plastic strain amplitude.
36:27
We track exactly how far the material is being physically
36:30
stretched and deformed beyond its elastic limit.
36:33
And there is a mathematical model for this similar to
36:36
Baskin. Yes, developed in the 1950s,
36:38
known as the Coffin Manson relationship.
36:41
Rather than using the materials innate strength, Coffin Manson
36:44
acts as a balance scale using the materials innate ductility.
36:48
OK, give me the spoken math for that one too.
36:50
Sure, it's delta epsilon sub b / 2 which gives us our plastic
36:54
strain amplitude, and that equals epsilon prime F which is
36:57
the fatigue ductility coefficient multiplied by the
37:00
quantity 2 times NF raised to the power of C.
37:03
On one side you have the physical plastic strain you are
37:06
inflicting. On the other side you have that
37:08
fatigue ductility coefficient. Essentially how much the metal
37:13
can stretch like taffy before tearing, scaled by the number of
37:16
cycles to failure. You've got it perfectly.
37:19
So we have two competing universes.
37:21
Baskin rules the high cycle universe, governed by the
37:24
strength of the metal Coffin. Manson rules the low cycle
37:27
universe governed by the ductility of the metal.
37:30
How do engineers bridge the gap? Through the strain life general
37:34
approach, it's wonderfully logical.
37:36
Total physical strain is simply the sum of elastic strain plus
37:40
plastic strain. Makes sense.
37:41
So the total strain amlitude equation just combines the two.
37:45
The elastic portion governed by Baskin added to the plastic
37:48
portion governed by Coffin Manson.
37:49
If you plot both of those curves on the same graph, do they
37:52
crossover? They do.
37:53
They intersect exactly what we call the transition life.
37:56
It's the physical point where the elastic damage and the
37:58
plastic damage are exactly equal.
38:00
What does that crossover point actually mean for an engineer
38:04
choosing a material to build a machine?
38:06
It represents A brutal engineering reality if your part
38:10
needs a long life. If you are operating well past
38:13
the crossover point in the high cycle regime, the elastic strain
38:16
is completely dominant. Therefore the strength of the
38:20
material controls its performance.
38:21
You want a hard, ultra strong metal to survive.
38:25
But if I'm building something that only needs to survive 100
38:28
massive violent impacts. Then you are operating before
38:31
the crossover point in the low cycle regime.
38:33
Here plastic strain dominates the damage.
38:37
The ductility of the material controls is performed
38:39
performance. You do not want a hard metal.
38:42
You want a soft, forgiving, ductile metal that can stretch
38:45
and absorb the energy without instantly tearing.
38:48
And here's where I see the cruelest irony in all of
38:50
material science. It is a profound irony.
38:53
In almost all engineering alloys, any thermal or chemical
38:56
process you use to increase the strength of the metal leads
38:59
directly and unavoidably to a reduction in its ductility.
39:03
And vice versa. You cannot have both.
39:05
You really can't. So if I take a piece of steel
39:07
and heat treat it, crunching it in oil to make it incredible?
39:10
It will perform Florida State in a high cycle fatigue environment
39:13
like a high speed motor shaft, but if you put that same
39:17
hardened steel into a low cycle environment with massive
39:20
yielding strains, it will shatter instantly.
39:23
Conversely, if you anneal the steel in a furnace to make it
39:26
soft and highly ductile, it survives low cycle fatigue
39:29
wonderfully, but fails rapidly under high cycle vibration.
39:34
You have to choose your weapon and your material based entirely
39:37
on the expected lifespan and operating environment of the
39:40
part. There is no master key.
39:42
That is incredible. OK so we know how cracks start,
39:46
we know how they grow and how the inherent properties of the
39:48
metal dictate its survival. But how do we actively fight
39:51
back? Like how do we stop the cracks
39:53
before they even begin? Let's talk about cheating
39:55
physics. How do engineers deliberately
39:58
pre stress materials to save them?
40:01
To understand how we cheat, you have to understand the
40:03
fundamental law of fracture mechanics.
40:06
Cracks generally will not initiate, nor will they
40:08
propagate in a compressively stressed zone.
40:10
Meaning what exactly? Meaning cracks need tension to
40:13
open up. Think about a crack in a piece
40:16
of plastic. If you pull the plastic apart,
40:19
the crack opens and tears further.
40:21
But if you squeeze the plastic together, the crack is forced
40:24
shut. It can't grow.
40:26
Oh. Obviously so.
40:28
The Holy Grail of fatigue prevention is to induce a
40:31
permanent state of residual compressive stress on the outer
40:35
surface of a component. So the goal is to permanently
40:37
squeeze the outside layer of the metal so that even when the
40:40
machine is running and pulling on the part, the net force at
40:43
the surface is still pushing inward, keeping the crack shut.
40:47
Exactly. You beat the tension by starting
40:49
with compression. A great real world example of
40:52
this something literally everyone listening interacts
40:54
with daily, is tempered glass, the screen on your smartphone or
40:58
the windshield of your car. How does tempering actually
41:01
create this residual squeeze? Tempering is a brilliant Thermo
41:05
mechanical process. Let's get a car windshield.
41:08
You take a sheet of ordinary brittle glass and heat it in the
41:10
furnace until it's soft, glowing and able to flow.
41:14
Then you blast both the outer surfaces with high pressure jets
41:17
of cool air. So the outside scenes of the
41:19
glass drop in temperature immediately.
41:21
Right, because they cool rapidly.
41:23
They contract, shrink and solidify into a hard, rigid
41:26
shell. But remember, the thick inner
41:29
core of the glass is still glowing hot and thermally
41:32
expanded. Right, the heat is trapped
41:34
inside. As the outer shell shrink, the
41:36
hot core just flows and squishes along with them, offering no
41:39
resistance. But eventually thermodynamics
41:42
wins. That hot inner core has to cool
41:45
down to room temperature too. And that is the trap.
41:48
When the inner core finally cools, it attempts to shrink and
41:51
contract, but it is now permanently fused to an outer
41:55
shell that is already cold rock hard and rigidly set in its
41:59
dimensions. So as the core shrinks, it pulls
42:02
fiercely inward on the outer shell.
42:04
It's dragging the surface inward, which puts the core into
42:07
a permanent state of tension, pulling itself apart, but it
42:10
puts the outer surface into massive permanent residual
42:13
compression. It is squeezing itself from the
42:15
inside out precisely. And this is vital because when
42:19
glass is loaded in tension, like when a rock hits your
42:22
windshield, failure almost always initiates from
42:25
microscopic flaws on the surface.
42:28
But for tempered glass, before the impact force could even
42:31
begin to exert a tearing opening load on those surface cracks, it
42:35
must first completely overcome that massive built in
42:39
compressive squeeze. It's like having a heavy duty
42:41
rubber band wrapped as tightly as possible around a block of
42:44
wood. Before you can pull the wood
42:46
apart, you first have to pull hard enough to defeat the
42:49
squeeze of the rubber band. That's why tempered glass can
42:52
sustain such massive impacts before shattering.
42:54
Yes, and we use the exact same concept on Metal, but obviously
42:58
we can't always heat in rapidly cool, complex massive metal
43:01
geometries like an engine block or a turbine blade the way we do
43:04
flat glass. Right, that would warp
43:05
everything. So we use a mechanical process,
43:08
We use a brute force. It's called shot.
43:09
Peening shot. Peening.
43:11
I absolutely love this concept. It sounds aggressive.
43:14
It literally involves bombarding a critical metal part with
43:17
millions of tiny spherical media called shot.
43:20
Like sandblasting, but instead of cutting sand you use little
43:24
metal, glass or ceramic balls. That's a perfect visualization.
43:28
You fire these tiny spheres at high velocity.
43:31
Each piece of shot striking the surface acts like a microscopic
43:35
high speed peening hammer. When it impacts the metal, it
43:39
imparts a tiny shallow indentation or dimple into the
43:42
surface. And what does that microscopic
43:44
dimple do to the metal underneath?
43:46
When the shot hits, it stretches the top layer of the metal
43:49
plastically. It yields the metal right at the
43:52
point of impact, forcing it to spread outwards like dropping a
43:54
bowling ball in mud. But the thick bulk of the
43:56
material underneath hasn't been hit.
43:58
It hasn't yielded. It is still perfectly elastic,
44:02
so the bulk underlying material acts like a giant spring trying
44:05
to push that stretched dimpled surface layer back to its
44:08
original size. So the top layer wants to be
44:10
bigger, the bottom layer is forcing it to stay small.
44:13
The net result? The net result is a highly
44:16
uniform, dense layer of material at the surface, locked in a
44:20
permanent, powerful state of residual compression.
44:23
It is identical to the tempered glass.
44:26
Typically, the residual compressive stress produced by
44:28
shot peening is at least half the yield strength of the
44:31
material being peened. Half the yield strength.
44:34
That's a massive invisible protective shield built right
44:37
into the metal. Where is this actually used in
44:39
the field? Everywhere that fatigue is the
44:41
primary mode of failure. Aerospace relies on it heavily.
44:45
The titanium blades, buckets and discs inside a jet engine are
44:49
intensely shot peened. Just to handle the vibrations.
44:52
Yeah, the blade routes where the attached to the hub are peened
44:55
to event fretting, galling and fatigue from the insane
44:59
rotational vibrations. You find it on the crankshafts
45:02
of heavy diesel trucks, automotive gear transmissions,
45:05
marine drive shaft. They even use a variant called
45:07
peen forming on the mats of aluminum wing skins of passenger
45:10
aircraft. By selectively bombarding one
45:13
side of the flat aluminum, they can gently curve the metal into
45:16
the aerodynamic shape of a wing while simultaneously
45:19
bulletproofing it against fatigue.
45:21
And hitting the metal with millions of tiny hammers doesn't
45:24
just create compression right? The metallurgical bonus effect.
45:27
Correct. It also provides intense cold
45:31
working of the surface. Just as we discussed earlier
45:34
with strain hardening, massive plastic deformation causes the
45:39
dislocations in the metal to tangle up and walk together.
45:42
So it's double protection. Yeah, this cold working
45:45
significantly hardens and strengthens the surface layer,
45:48
providing drastically better resistance against wear,
45:51
intergranular corrosion, and it even physically hammers shut
45:55
microscopic porosity in cast metal parts.
45:58
It is just a brilliantly elegant engineering solution.
46:01
Use brute force to induce a squeeze that stops cracks from
46:05
breathing. Now we've been talking about
46:07
macroscopic structures, jet engine blades, car windshields,
46:10
suspension bridges. The big stuff.
46:12
Yeah, but to truly understand material failure, I want to zoom
46:15
all the way in, past the dimples, past the surface.
46:18
What does a crack actually look like when it travels through the
46:21
atomic lattice itself? Let's dive into the microscopic
46:24
battlefield. To set the stage here, we need
46:26
to correct a common misconception.
46:28
Metals are not just uniform, homogeneous blocks of matter.
46:32
They are crystalline. OK.
46:34
When liquid metal cools and solidifies in a foundry, it
46:37
forms microscopic crystals called grains.
46:41
These grains grow outward until they bump into one another.
46:44
The borders where the different crystals meet are called grain
46:46
boundaries. OK, so let's set up a
46:48
theoretical scenario based on the latest metallurgical
46:51
research. We have two samples of the exact
46:54
same metal alloy. The 1st is what we call a
46:57
nanocrystalline material. The average grain size is
47:00
incredibly small, almost unimaginably small, say just 40
47:04
nanometers across. Very tiny.
47:06
The second sample is a microcrystalline material where
47:08
the grains are much larger, say larger than one micrometer.
47:12
That's a huge difference in scale.
47:14
How does a fatigue crack behave differently as it tries to tear
47:17
through each of these structures?
47:18
The difference is striking and it dictates the toughness of the
47:21
metal. It all comes down to how the
47:23
crack tip interacts with those grain boundaries in the
47:26
nanocrystalline material. With the ultra tiny 40 nanometer
47:31
grains, the path of the crack under simple cyclic tension is
47:36
perfectly linear. It is undeflected.
47:38
It acts like a laser beam. It just cuts straight through.
47:40
Why doesn't hit the boundaries? It does, but the grains are so
47:45
unimaginably tiny relative to the plastic zone at the tip of
47:48
the crack that the material effectively acts as a
47:51
homogeneous, continuous solid. The crack simply slices straight
47:55
through the macro structure. It doesn't see the individual
47:58
grains as structural obstacles because they're just too small
48:00
to deflect the massive energy of the crack tip.
48:03
But what happens when we move to the microcrystalline regime
48:06
where the grains are 50 or 100 times larger?
48:08
Now the crack encounters massive distinct crystal boundaries.
48:13
And remember, each grain has its atonic lattice oriented in a
48:16
slightly different direction. Oh, so the grid changes
48:18
direction? Exactly.
48:20
When the crack reaches a grain boundary, it can't just slice
48:23
straight across. The weakest atomic planes in the
48:26
new grain are pointing in a different direction.
48:28
The crack has to physically reorient itself to follow the
48:32
path of least resistance through the new crystal lattice.
48:35
So the crack is forced to turn. This is what researchers call
48:38
crystallographic crack path deflection.
48:41
Exactly. The crack literally zigs and
48:44
zags. It tilts and twists 3
48:47
dimensionally as it navigates the maze of crystal boundaries.
48:50
Now I can visualize that, but does this actually slow the
48:53
crack down? I mean, geometrically it has to
48:56
travel a slightly longer total distance because it's zigzagging
48:59
instead of going in a straight line, but is there a physical
49:01
mechanical benefit to the deflection?
49:04
Yes, a massive 1. It's not just about the
49:07
distance. When the crack deflects and
49:09
tilts away from the primary plane of stress, the effective
49:12
stress intensity at the very tip of the crack is mathematically
49:16
reduced. Let me make sure I understand
49:18
because it turned a corner at the main force.
49:19
Pulling the part apart isn't pulling directly on the tip
49:22
anymore. Exactly.
49:24
The force driving it forward is diluted because the tip is no
49:27
longer pointing directly into the main tensile load.
49:30
It's pulling at an angle. It loses efficiency.
49:33
That's incredible. So a larger grain structure is
49:36
actually a natural defense mechanism against crack growth.
49:39
It is, and the zigzag in geometry induces A secondary
49:42
defense mechanism known as roughness induced crack closure.
49:46
Roughness induced closure. What does that mean physically?
49:50
Because the crack path is so jagged and tortuous, the two
49:53
faces of the crack aren't perfectly smooth mirror images
49:56
anymore. During the unloading part of the
49:58
fatigue cycle, as the metal tries to relax, those jagged
50:02
mountainous edges actually rub against each other.
50:04
They wedge together behind the crack tip, effectively propping
50:07
the crack open. Wow, so the jagged edges act
50:10
like a doorstop. Yes, they physically shield the
50:12
tip of the crack from experiencing the full range of
50:14
stress during the next cycle. It absorbs the energy, so larger
50:17
grain structures in the specific context of crack propagation,
50:21
physically slow down fatigue failure by forcing the crack to
50:24
take a torturous energy draining path and waging itself shut.
50:27
I absolutely love that visualization.
50:30
A crack desperately trying to navigate a microscopic maze of
50:34
crystal boundaries, losing momentum and energy with every
50:37
single turn. That's.
50:38
A great image. But as fascinating as that is,
50:41
to truly predict these failures in the real world, especially
50:44
when we push materials into extreme disaster scenarios where
50:47
they are yielding, stretching, and tearing on a massive scale,
50:51
we need the heaviest mathematical artillery in all of
50:54
engineering. And that brings us to our final
50:56
topic, elastic plastic fracture mechanics and the infamous J
51:01
integral. To truly appreciate the genius
51:04
of the J integral, we have to look back at the historical
51:06
limitations of early fracture mechanics.
51:09
In the mid 20th century we had the Williams solution and linear
51:12
elastic fraction mechanics. These relied on a parameter
51:15
called K, the stress intensity factor.
51:18
But K came with a massive restrictive assumption.
51:21
Small scale yielding, Small scale yielding meaning the
51:24
little zone of plastic permanent deformation right at the very
51:28
tip of the crack has to be absolutely tiny compared to the
51:31
overall size of the steel beam or part.
51:34
The vast majority of the part has to remain perfectly springy
51:37
and elastic. Right.
51:38
If the part is massive and the crack is tiny, Kay works
51:42
beautifully, right? But what happens in a real world
51:45
catastrophic failure? Or what happens when you are
51:48
testing modern, highly ductile, super tough metals?
51:52
That plastic zone isn't tiny it. Can be massive.
51:55
The metal is visibly stretching, flowing and tearing far beyond
51:58
its elastic limits long before it finally snaps.
52:01
When this large scale yielding happens, the linear equations
52:04
behind K completely collapse. They produce garbage data.
52:07
Enter Professor Jr. Rice, 1968.
52:10
Yes, Professor Jr. Rice, alongside brilliant
52:13
related work by researchers like Chairpenov, Eshleby and Sanders,
52:16
revolutionized the entire field by introducing the J interval.
52:20
It is a mathematical and thermodynamic tool specifically
52:23
designed for situations where there is massive large scale
52:25
yielding. Let's unpack this conceptually.
52:28
How does it work? I know from the literature it
52:30
involves drawing a mathematical path or a contour loop around
52:34
the crack tip. Exactly.
52:36
The J integral is a path independent line integral.
52:39
Mathematically, you draw a line starting from the lower face of
52:42
the crack, looping counterclockwise entirely around
52:46
the chaotic plastic zone at the crack tip, and ending on the
52:49
upper face of the crack. In simple physical terms, what
52:53
is this integral actually measuring inside that loop?
52:56
Physically, it represents the rate of change of the net
52:59
mechanical potential energy with respect to crack advance for a
53:03
nonlinear elastic solid. But to put it much more simply,
53:07
it is a direct measure of the energy flowing into the crack
53:10
tip from the surrounding material.
53:12
So it's tracking the energy. It tells you exactly how much
53:14
juice is driving the crack forward, even when the
53:18
surrounding metal is permanently yielding and warping.
53:20
You said earlier it's path independent.
53:22
Every paper on the subject highlights this.
53:24
This is the superpower of the J integral.
53:27
But what does path independence actually mean for an engineer
53:30
standing in a testing lab? It is nothing short of a
53:32
mathematical miracle. Path independence means that the
53:36
equation yields the exact same answer regardless of how close
53:40
or how far away you draw your integration path from the actual
53:44
crack tip. Wait, really?
53:46
Explain how we use that. Let's say you have path one.
53:48
It tightly hugs the microscopic tip of the crack right in the
53:51
heart of the plastic zone where everything is chaotic, atoms are
53:54
sliding, heat is being generated, and it is physically
53:58
impossible to attach sensors to measure the stress accurately.
54:01
OK, an impossible measurement. Then you draw path to.
54:04
Path to is a massive loop drawn far away from the crack way out
54:07
on the outer edges of the metal block in the elastic region
54:10
where the stresses and strains are smooth, predictable, and
54:13
incredibly easy to measure with standard cheap strain gauges.
54:17
And the math of the J integral guarantees that the energy
54:20
flowing through the massive, easy to measure path 2 is
54:24
exactly equal to the energy flowing into the microscopic
54:26
chaotic path 1. Exactly the energy is conserved
54:29
within the contour J calculated along path 1 equals J calculated
54:35
along path 2. This allowed engineers for the
54:38
first time in history, to accurately determine the
54:40
catastrophic conditions at the microscopic crack tip by taking
54:44
simple large scale measurements far away from the tip.
54:47
That is utterly brilliant. It's like calculating exactly
54:50
how much water is draining out of a microscopic hole in a sink
54:53
by measuring the gentle flow of water at the edges of the basin.
54:57
Now this doesn't just sit in a vacuum, it connects back to the
55:00
older linear models, doesn't it? Yes, it is fully backward
55:03
compatible. The GNTR goal is formulated
55:05
using deformation theory of plasticity, which is essentially
55:07
equivalent to nonlinear elasticity.
55:10
Because of this, it forms a perfect bridge for the special
55:13
case of a perfectly brittle linear elastic solid like our
55:16
chalk or high strength glass. J is mathematically exactly
55:20
equal to G the strain energy release rate.
55:23
Why is that equivalent so important practically?
55:25
I mean, why did this save the testing industry?
55:28
Because of test specimen size requirements to accurately
55:31
measure the fracture toughness of a highly ductile metal using
55:35
the old linear K methods, the standards dictate that the
55:38
specimen has to be unimaginably massive to enforce that small
55:43
scale yielding rule. How?
55:45
Massive are we talking? If you wanted to test the
55:47
toughness of a low strength steel used in a submarine hall
55:51
using linear methods, the required solid block of steel
55:54
might need to be over a meter thick.
55:56
A meter thick. You'd need a testing machine the
55:59
size of a building just to pull it apart.
56:00
It's physically impossible and absurdly expensive.
56:03
Exactly. The industry was hitting a wall,
56:07
but with the J integral. Because J can handle large scale
56:09
yielding without breaking a sweat.
56:11
Exactly. The specimen size requirements
56:13
for AJ test are exponentially less severe.
56:16
You can take a highly ductile steel machine, a small deeply
56:19
cracked compact tension specimen, maybe just an inch
56:22
thick, put it in a standard lab machine, measure the critical J
56:26
value at failure, and then use the mathematical equivalents to
56:28
back calculate what the equivalent linear toughness
56:31
value would be. That's.
56:32
Genius. It literally saved the entire
56:35
field of fracture testing from structural and economic
56:39
impossibility. Wow, we went from requiring a
56:43
block of steel the size of a refrigerator to something you
56:46
can hold in your hand, all because Jr.
56:47
Rice proved the energy flow is path independent.
56:51
OK, one final concept here to wrap up the physics.
56:54
If the metal is stretching so much under this large scale,
56:57
yielding, if the J integral is working so hard, the crack
57:00
itself isn't just a microscopic mathematically sharp line
57:03
anymore, right? It must physically deform.
57:06
Absolutely. The mathematical concept of a
57:08
sharp crack tip breaks down in reality, and this brings us to a
57:11
physical measurement called CTO, the crack tip opening
57:14
displacement. Crack tip opening displacement.
57:17
Yeah, because of the massive plastic deformation happening at
57:20
the tip, the crack literally blunts.
57:22
Before the crack actually advances forward and tears new
57:25
metal. The tip stretches and opens up
57:27
into AU shape. So you can physically measure
57:29
how far the jaws of the cracks stretch open before they snap.
57:32
Yes, the strict definition is slightly arbitrary, since the
57:36
opening width varies as you approach the exact curved tip.
57:40
So engineers use an operational definition, often based on a 45°
57:44
construction line drawn from the crack tip back to the crack
57:47
faces. But the critical take away is
57:49
that this CTOD is used as a direct physical measure of the
57:53
material's toughness. Like a visible gauge?
57:56
Exactly. It is an experimental visual
57:58
measure of the driving force behind the fracture.
58:01
The more a crap tip can blunt, stretch and yawn open without
58:05
actually advancing forward, the tougher and safer the material
58:08
is. It's the material desperately
58:10
trying to absorb the energy by stretching its jaws rather than
58:13
letting the crack RIP forward. This has been absolutely
58:16
phenomenal, from the macro world of airplane wings down to the
58:19
atomic lattice. As we wrap up this master class,
58:21
do you have a final provocative thought to leave the listener
58:24
with? I do think about everything
58:27
we've covered today, from the tiny invisible hydrogen atom
58:30
sneaking in and embrittling a steel pipeline, to the
58:33
zigzagging tortuous path of a crack navigating A
58:36
microcrystalline lattice, all the way to the mathematically
58:40
perfect path independent J integral capturing the massive
58:44
energy flow of yielding steel. It's a lot to take in.
58:46
It is our physical world. The materials we build our
58:49
civilization with are a constant, invisible battleground
58:53
of energy. The next time you look at a
58:55
suspension bridge, a skyscraper, or the spinning jet engine on an
58:58
aircraft, don't look at it as a piece of static dead metal.
59:03
Look at it as a living, dynamic system.
59:05
It is constantly fighting against the microscopic forces
59:08
of fatigue, desperately holding onto its residual compressive
59:11
stresses, relying on its grain boundaries to deflect damage
59:14
just to survive one more cycle. What a profound perspective.
59:18
Remember, design junkies master the critical few or the trivial
59:21
many will master you. If you're leaving this episode
59:23
with something you didn't and arrive with, drop a comment.
59:26
Takes you a few seconds, but it tells me this workbench is worth
59:28
keeping open.