Latest / Mechanical Engineering Made Simple / (#115) How Torsion and Fatigue Break Mechanical Shafts.

Transcript

(#115) How Torsion and Fatigue Break Mechanical Shafts. · 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
0:08
below. Imagine for a second that you
0:10
were staring at a massively expensive jet engine.
0:14
I mean, it was Florida State designed in CAD.
0:16
You know, the mathematical model said it was perfectly safe.
0:19
Right, the materials were pristine when they rolled off
0:21
the assembly line. Exactly.
0:23
But then suddenly at 30,000 feet a critical shaft just it snaps
0:27
in half and the engine rips itself apart.
0:29
Like why did it fail? Yeah, how does a design that
0:32
works perfectly on a computer screen become an absolute
0:35
disaster in the real world? So today we are taking a massive
0:39
deep dive into the most foundational yet complex
0:42
questions that you, the engineer, asked about mechanical
0:46
design, stress materials, and mathematical optimization.
0:50
Because the gap between a digital model and physical
0:53
reality, well, that's where the tree art of engineering lives.
0:56
I mean, we have a mountain of data today.
0:58
Oh for sure, textbooks full of stress analysis algorithms.
1:01
Yeah, and vast libraries of material science data, plus
1:06
structural optimization software that runs like millions of
1:11
calculations a second. But looking up a yield strength
1:14
number in a handbook, that doesn't make you a great
1:16
designer. Yeah, it really doesn't.
1:18
The true skill is connecting the docs.
1:20
It's understanding the fundamental micro mechanics of
1:22
how a part fails. You know how the specific
1:25
crystalline structure resists that failure, and how we
1:28
mathematically optimize the geometry so it survives without
1:32
over building it into some massive heavy block of metal?
1:35
OK, let's unpack this. Because the reality of
1:37
engineering is that designing a part is, well, it's relatively
1:40
easy. Drawing a beautiful 3D model on
1:42
your computer is easy. Oh yeah, anyone can make
1:45
something look good and CAD. But designing a part that
1:48
actually survives the chaotic, unpredictable real world?
1:51
That is an entirely different beast.
1:53
It really is. I mean, you might think you've
1:55
designed the perfect rotating shaft for an industrial motor.
1:58
But before we can even begin to talk about things breaking, we
2:01
have to understand the invisible mathematical forces acting upon
2:05
them. We can't see stress, right?
2:07
We can't see strain. Exactly.
2:08
We have to model it, which raises a huge question for
2:12
anyone designing rotating machinery.
2:14
How do we accurately quantify the twisting and bending forces
2:19
tearing at our designs? Yeah.
2:21
Like how do we put a number on that?
2:22
Well, to quantify it, we really have to visualize the
2:24
microscopic reality of the material.
2:27
Let's let's look at the anatomy of stress, specifically
2:31
torsional stress in circular shafts.
2:33
OK, so twisting a metal bar, right?
2:35
When you hold a metal bar fast at one end and apply a twisting
2:39
force, a torque to the other, you aren't just stretching the
2:42
metal. Every single infinitesimally
2:45
thin circular cross section of that bar is actively trying to
2:48
slide over the one next to it. It's trying to shear off.
2:51
Exactly. Imagine a stack of thousands of
2:53
thin metal coins glued together. Oh, that's a good way to
2:55
picture. It right, so when you twist the
2:57
stack, the glue between every single coin is being pushed to
3:01
its limit as each crane tries to rotate past its neighbor.
3:04
And if you were to draw a perfectly straight line down the
3:06
side of that shaft with a marker before you twisted it, that line
3:10
wouldn't just stretch. No.
3:11
You would wrap around the cylinder and become a continuous
3:14
Helix. The angle of that twist
3:16
accumulates over the length of the bar.
3:18
And here is the critical part for you as a designer.
3:21
The shearing stress at any given point inside that material is
3:25
directly proportional to its radial distance from the central
3:29
axis. Right, which means the material
3:31
at the dead center of the shaft, the neutral axis, isn't moving
3:34
relative to itself. It's experiencing essentially 0
3:37
torsional stress. Wow, zero stress the center, but
3:40
the material at the absolute outer surface, like the skin of
3:43
the shaft that's taking the maximum beating.
3:46
Absolutely. It is being stressed, touched
3:48
and sheared to its absolute limit.
3:50
And that physical reality drives one of the most fundamental
3:53
geometry choices in mechanical engineering.
3:56
You're talking about the solid versus the hollow shaft, right?
3:58
Yep, because the material in the center of a solid shaft is
4:02
experiencing 0 stress, It's doing practically 0 work to
4:07
resist the twisting moment. It's just dead weight.
4:10
Exactly. It's essentially dead weight,
4:12
and if you are designing for aerospace or automotive or
4:15
anywhere weight matters, that dead weight is unacceptable.
4:18
So hollow shafts are vastly more efficient in torsion them.
4:21
Incredibly so. By removing the inner core and
4:24
pushing that same massive material out to a larger outer
4:27
diameter, you drastically increase what we call the polar
4:31
moment of inertia. So you're putting the metal
4:33
exactly where the stress is highest.
4:35
Exactly. You put it where it's needed
4:37
most. Which is exactly why biological
4:39
structures evolved this way. I mean look at bird bones or
4:41
human femurs or even engineered structures like the massive Dr.
4:46
Shafts underneath semi trucks. Right, they are all hollow
4:49
tubes. They maximize torsional
4:51
stiffness while shedding pounds. But you know, as an engineer, I
4:54
can't just say make it hollow, I need a hard number.
4:57
How do we actually calculate that outer maximum stress?
5:01
Well, there is a very specific mathematical relationship
5:04
engineers rely on when sizing shafts for transmitting power.
5:08
OK, let's hear the math. When a solid round shaft is
5:11
transmitting power from a motor to a load, we can calculate that
5:14
maximum unit shearing stress directly, right?
5:17
The formula states that Tau equals 321,000 * H divided by
5:22
the product of N and d ^3. OK, so Tau equals 321,000 * H
5:28
divided by the product of N&D cubed.
5:30
What do those variables mean in the real world?
5:33
So Tau represents the maximum shear stress, H is the
5:36
horsepower being transmitted by the system, N is the rotational
5:40
speed and revolutions per minute, and D is the diameter of
5:43
the shaft. Got it.
5:44
So it's an inverse cubic relationship with the diameter.
5:46
If I make the shaft just a tiny bit wider, the stress drops
5:50
massively because of that D cubed term in the denominator.
5:54
Exactly. The diameter has a massive
5:56
impact, so you plug in your motors horsepower, you plug in
5:58
the RPM and you immediately know the exact shear stress trying to
6:02
RIP the outer skin of your shaft apart.
6:04
But there's a massive catch here, isn't there?
6:06
I mean, in the real world, a pure isolated state of torsion
6:09
almost never exists. It is incredibly rare.
6:13
Real parts in real machines experience complex combined
6:17
stresses. Think about that rotating motor
6:20
shaft again. Right, it's not just floating in
6:22
space. Exactly, it isn't twisting
6:24
freely in a vacuum. Maybe there is a heavy Dr.
6:27
pulley mounted on the end of it, pulling down with hundreds of
6:30
pounds of belt tension. Or maybe there's a helical gear
6:33
meshing with another gear pushing radially and axially
6:36
against the shaft. Yep, that means the shaft is
6:38
bending, so it's experiencing longitudinal tension and
6:42
compression at the exact same time it is experiencing
6:46
torsional shear. And these forces don't politely
6:49
take turns, they stack right on top of each other.
6:51
They absolutely do. The longitudinal normal stresses
6:54
from bending interact continuously with the shearing
6:57
stresses from Georgian. As a designer, you have to
6:59
figure out the worst case scenario happening inside that
7:02
three-dimensional stress state. And this is where Moore Circle
7:05
becomes an absolutely vital non negotiable tool, doesn't it?
7:08
Oh 100% Moore circle is essential here.
7:11
You know, Moore Circle was always one of those things that
7:13
looked completely terrifying on a chalkboard in engineering
7:17
school. Oh, I know it scares a lot of
7:18
students. But conceptually, I like to
7:20
think of Moore Circle like a camera.
7:23
Imagine a camera orbiting around a microscopic 3 dimensional cube
7:27
of metal hidden and deep inside the shaft.
7:30
That's a really great way to visualize it.
7:32
Right, and as that imaginary camera changes its viewing angle
7:35
rotating around the cube, the mix of pulling forces and
7:39
sliding forces changes on the screen.
7:42
Exactly. Moore Circle is essentially just
7:44
a graphical and mathematical method of finding the exact
7:47
camera angle where the pulling force, the normal stress, looks
7:51
the absolute biggest and the sliding force, the shear stress,
7:55
completely disappears. And we call those the principal
7:57
planes and the principal stress stresses.
7:59
That orbital camera analogy perfectly captures the
8:02
mathematics of coordinate transformation.
8:04
So you're just hunting for the worst possible angle.
8:07
Exactly. You are rotating your frame of
8:09
reference until you find the plane of maximum normal stress,
8:12
or conversely, the plane of maximum shear stress.
8:15
Because the material doesn't care what arbitrary XYZ
8:19
coordinate system you drew in your CAD program.
8:21
Right. The material only feels the
8:23
maximum principal stress acting on its internal crystal lattice.
8:28
So if a shaft is simultaneously twisting and bending, how does
8:32
the metal actually know when to yield?
8:35
Like how does it know when the combination of these forces is
8:38
just too much? Well, that's where failure
8:40
theories come in. I always think of the analogy of
8:42
ringing out a wet towel. If you take a wet towel and just
8:45
pull it straight from both ends, pure tension.
8:48
It takes an immense amount of force to actually RIP the
8:51
fabric. It does.
8:52
It's surprisingly strong in pure tension.
8:54
But if you ring it out, twisting it tightly while simultaneously
8:57
pulling it from both ends, you are combining severe shear with
9:01
tension. The fibers of the towel are
9:03
going to snap much, much sooner than they would in pure tension
9:06
alone. That's right, the towel knows to
9:09
yield because those combined stresses amplify each other on
9:12
specific microscopic planes. The towel analogy hits on the
9:15
exact reason engineers have spent over a century developing
9:19
specific failure theories. We need to predict exactly when
9:23
that towel will snap based on its specific material
9:27
properties. And different materials act
9:28
completely differently, right? Yes.
9:30
The crucial realization is that different materials know they're
9:34
failing in fundamentally different ways.
9:36
We really have to contrast the criteria we use for brittle
9:39
materials versus ductile materials.
9:41
OK, let's start with brittle materials like cast iron, glass,
9:45
ceramics, or even concrete. For those, we rely on the
9:48
maximum stress theory, which is historically called Rankine's
9:51
theory. Ranking's theory.
9:53
What does that assume? This theory assumes that a
9:55
brittle material will fail the exact moment the maximum
9:58
principal tensile stress reaches the ultimate tensile strength of
10:02
the material. So it completely ignores the
10:04
other axis. Exactly.
10:06
It operates. It's completely regardless of
10:07
what the sheer stresses or the compressive stresses are doing
10:10
on the other axis. Brittle materials don't flow,
10:14
they fail by sudden catastrophic cleavage.
10:16
The atomic bond simply pull apart.
10:18
Like snapping a piece of blackboard chalk in your hands.
10:21
There's no warning, no stretching, just a sudden pop
10:24
and it's in two pieces. Precisely the behavior.
10:27
But if you are designing with ductile materials like
10:30
structural steel, aluminum or brass strove, they behave
10:34
completely differently. They yield.
10:36
Right. They bend before they break.
10:38
Yes, they deform plastically and permanently before they ever
10:41
reach their breaking point. For ductile models, we use the
10:44
distortion energy theory. The famous von Mises criterion.
10:48
Every mechanical engineer knows that name.
10:50
Yep, von Mises and this theory is much more sophisticated.
10:54
It states that yielding begins when the distortion strain
10:57
energy per unit volume reaches the distortion strain energy at
11:01
yield in simple tension. OK, let's break that down
11:03
because distortion strain energy is a pretty heavy term.
11:06
What is actually happening to the metal physically?
11:09
Think of it this way. Any complex stress state can be
11:12
broken down into two components. First, there is hydrostatic
11:16
stress, which is uniform pressure from all sides.
11:19
Like dropping a block of steel to the bottom of the Mariana
11:22
Trench. Exactly.
11:23
It experiences immense hydrostatic pressure.
11:26
It will shrink slightly in volume, but it won't yield or
11:29
change its overall shape because the pressure is equal
11:32
everywhere. OK.
11:33
So that's hydrostatic. What's the second component?
11:36
The second component is deviatoric stress, or distortion
11:39
stress. This is the stress that actually
11:41
tries to change the shape of the cube, stretching it into a
11:44
rectangle or skewing it into a parallelogram.
11:47
Oh I see, so it's the shape changing force.
11:49
Right, the von Mises criterion calculates the energy of just
11:53
that shape changing distortion. It takes all your complex 3D
11:56
stresses, strips away the harmless hydrostatic pressure,
12:00
and calculates a single equivalent stress.
12:03
And then you just compare the equivalent von Mises stress to
12:06
the standard yield strength you looked up in your textbook.
12:08
Exactly. It is an incredibly accurate
12:11
predictor for how the crystalline lattice of a ductile
12:13
metal decides to slip and flow. OK, so we've done the math.
12:17
We've calculated the static stresses using the torque and
12:20
bending loads, we rotated our camera with more circle, we
12:24
calculated the von Mises equivalent stress, checked our
12:27
material properties, and the math says the part should
12:30
theoretically survive with a massive factor of safety.
12:33
Right, the math checks out perfectly.
12:35
We build it, we turn the machine on, it runs Florida State.
12:39
So why does that exact same shaft suddenly snap in half
12:43
after a year of use under loads we know are completely safe and
12:46
well below the yield strength? Well, that question transitions
12:50
us out of the comforting, predictable world of static
12:52
loading, and it drops us straight into the chaotic,
12:55
dynamic world of fatigue and fracture.
12:57
Ah, fatigue, the silent killer. Exactly what you are describing
13:02
is a progressive failure phenomenon.
13:04
Fatigue happens when a material is subjected to cyclic loading
13:07
forces that fluctuate, reverse or pulse over time.
13:11
So it's the repetition that kills it.
13:13
Right. The insidious nature of fatigue
13:15
is that even if your peak operating stress is vastly below
13:19
the von Mises yield strength of the material, the repeated
13:22
thousands upon thousands of applications of that low stress
13:26
can initiate highly localized microscopic damage.
13:30
So like a tiny slip band forms on the surface?
13:33
Yep. Which turns into a micro crack
13:35
which slowly grows with every single revolution of the motor
13:38
until the remaining cross section of the shaft is too
13:41
small to carry the load and it just snaps.
13:43
The classic example of this is bending a paper clip right.
13:46
You can't pull a paper clip apart with your bare hands, it's
13:49
too strong and static tension. No, you'd cut your finger years
13:52
before it broke in pure tension. But if you bend it back and
13:54
forth a dozen times, it easily snaps.
13:57
We generally divide this fatigue behavior into two main regimes,
13:59
right high cycle fatigue and low cycle fatigue.
14:02
We do, and the distinction is vital for how we design.
14:06
High cycle fatigue involves relatively low fluctuating loads
14:09
where the global strain in the material remains almost entirely
14:13
elastic. So the part springs back to its
14:14
original shape every single time.
14:16
Exactly because the plastic damage per cycle is
14:20
infinitesimally small, may be localized to a few grains of
14:24
metal at a surface scratch, it typically takes a massive number
14:28
of cycles to cause failure. Like how massive are we talking?
14:31
Usually defined as greater than 10,000 cycles, but often
14:34
stretching into the millions or billions for things like engine
14:37
valves. Wow, billions of cycles and low
14:40
cycle fatigue is the opposite. Right Low cycle fatigue involves
14:43
massive loads that push significant portions of the
14:46
material into the plastic strain range on every single cycle, so.
14:50
It's permanently deforming a little bit every time, yeah.
14:53
The damage accumulates brutally fast, leading to catastrophic
14:57
failure in fewer than 10,000 cycles.
14:59
Here's where it gets really interesting, because if we look
15:02
at the real world, a part isn't just sitting in a laboratory
15:05
testing machine experiencing a single perfect constant sine
15:09
wave of repeating load. No, real world loads are a mess.
15:12
Think about the suspension control arm on your car.
15:15
You back out of your driveway, which is a small load.
15:18
Then you drive on a perfectly smooth highway for an hour,
15:21
accumulating thousands of tiny stress cycles.
15:23
Right Then you hit a massive pothole, which is a huge stress
15:26
spike. Then you drive on a violently
15:29
corrugated gravel Rd. The stress spectrum is complete
15:33
chaos. How on earth do you account for
15:35
a history of varying loads? That is one of the most
15:38
difficult challenges in mechanical design, honestly, and
15:41
we handle it using cumulative damage theories.
15:44
OK, cumulative damage. What's the main theory there?
15:46
The most famous and widely utilized is the Ponggren Minor
15:50
Linear Damage rule. It is a brilliant, highly
15:53
intuitive concept. It postulates that every single
15:57
stress cycle, no matter how small, consumes a specific
16:00
fraction of the materials total fatigue life.
16:03
I've always thought of the Palmgren minor rule like a
16:05
bucket of structural health. Oh, that's a good analogy.
16:08
Yeah, when the part rolls off the assembly line, the bucket is
16:10
completely full of water, and every time you apply a load
16:13
cycle, you scoop a little water out.
16:15
Exactly the right visualization. If you look at an SN curve, a
16:19
graph of stress versus number of cycles to failure, you might see
16:23
that at a specific high stress level, the part will fail in
16:26
exactly 100,000 cycles. If you drive over that
16:29
corrugated gravel road and subject the suspension to 10,000
16:33
cycles at that high stress, you have scooped out 10% of the
16:38
water, you have consumed 10% of the parts life.
16:41
And miners rule just lets you add those up.
16:43
Yes, miners rule allows you to linearly add up these damage
16:47
fractions from completely different load spectrums.
16:50
The driveway, the highway, the pothole.
16:52
You calculate the ratio of cycles applied divided by cycles
16:55
to failure at every stress level.
16:57
So when the sum of those fractions reaches 1.0 or 100%,
17:01
the bucket is empty and the math predicts failure.
17:04
You've got it, but there is an issue with the palm gun miner
17:08
rule. What's that?
17:09
Well, it and SM curves in general are fundamentally
17:12
assuming we're starting with a perfect part.
17:14
They track the initiation of a crack and its eventual failure.
17:18
Right. But let's be totally honest
17:20
about manufacturing. Perfect materials do not exist.
17:23
There are inherent flaws in everything.
17:25
Absolutely. Everything.
17:26
There are micro inclusions of slag from the casting process,
17:29
microscopic scratches from the lathe cutting tool, tiny voids
17:33
from welding. What if the crack is already
17:35
there on day one? That realization that nothing is
17:39
perfect forms the entire basis of fracture mechanics.
17:42
It represents a massive philosophical shift in
17:45
engineering. How so?
17:47
We stop asking the traditional question of when will a crack
17:49
form and we just accept reality. We assume flaws already exist.
17:54
So what's a new question then? The new questions become how
17:57
fast will this existing flaw grow under my cyclic loads, and
18:01
at what exact length will it cause a sudden explosive
18:04
fracture? So we abandoned simple stress
18:07
calculations entirely. Pretty much we compare 2 highly
18:10
specific key parameters instead, the stress intensity factor and
18:14
the fracture toughness. OK, let's define those because
18:17
they sound similar, but they do very different things, right?
18:19
Very different. The stress intensity factor,
18:21
denoted by the letter K characterizes the strength of
18:24
the stress field acting right into the unimaginably sharp tip
18:27
of the crack. So it's an applied condition.
18:29
Yes, it depends on the global load you are climbing to the
18:33
part, the geometry of the part, and critically, the current
18:36
length of the crack. And what about fracture
18:38
toughness? Fracture toughness, on the other
18:40
hand, is a fundamental inherent property of the material itself.
18:44
Just like density or melting point, it is the critical
18:47
maximum value of that stress intensity factor that the atomic
18:51
lattice can actually withstand. So if your applied stress
18:54
intensity factor reaches the materials fracture toughness,
18:57
the crack abruptly becomes unstable.
18:59
Exactly. Stops growing slowly and
19:01
propagates at nearly the speed of sound, shattering the part
19:05
instantly. And to track how that crack
19:06
grows over time, cycle by cycle, creeping closer to that critical
19:10
fracture, toughness engineers use a mathematical relationship
19:14
called the Paris law. Yes, the Paris law is crucial
19:17
here. Using spoken math, the Paris law
19:19
states the derivative of A with respect to n = C multiplied by
19:24
the quantity delta K raised to the NTH power.
19:28
Let's translate that into physical reality.
19:30
Sure. So a is the length of the crack,
19:31
N is the number of fatigue cycles.
19:33
So data N is simply the rate of crack growth per cycle.
19:37
Right. Delta K is the range of the
19:39
stress intensity factor during that cycle, and CNN are
19:43
constants specific to the metal you're using.
19:45
What the Paris law reveals is honestly terrifying for
19:49
longevity. It tells us that crack growth is
19:51
not linear, it's exponential. It absolutely is, because the
19:55
stress intensity factor depends on the crack length.
19:58
As the crack gets just a tiny bit longer today, the stress
20:01
concentration at its tip increases.
20:03
That acts like a magnifying glass, focusing the load tighter
20:07
and tighter. Exactly.
20:08
Which means tomorrow the crack grows faster than it did today,
20:12
and the day after that it grows even faster.
20:15
The Paris Law models an accelerating death spiral for
20:18
the component. But there is a limit to the
20:20
math, right? I mean, the Paris law and linear
20:22
elastic fracture mechanics work beautifully on brittle
20:24
materials. Or when the crack is really
20:26
small. They do, but they assume the
20:28
material right at the incredibly sharp tip of the crack is
20:31
behaving elastically in ductile metals.
20:34
That just isn't true. The stress gets so insanely high
20:37
at that microscopic point that the metal yields.
20:40
Right. A tiny bubble of plastic
20:41
deformation forms right at the crack tip.
20:44
Yes, the crack tip plastic zone. And when that plastic zone gets
20:48
too large compared to the thickness of the part, the
20:51
linear math completely breaks down.
20:53
It gives you dangerously inaccurate predictions.
20:56
That is exactly when structural integrity engineers must shift
20:59
their methodology. We have to use failure
21:02
Assessment diagrams or FA DS. FA DS, How do those work?
21:06
FED is a highly sophisticated 2 dimensional plot that
21:10
acknowledges a part can fail in 2 completely different ways
21:12
simultaneously. It assesses the risk of brittle
21:15
fracture on the Y axis and plastic collapse which is gross
21:19
yielding of the entire remaining cross section on the X axis.
21:22
So it's plotting both risks at once.
21:24
Yes, the diagram has a failure curve drawn on it.
21:27
You take your operating conditions, your crack size, and
21:29
your loads and you plot them as a single coordinate point on
21:32
this diagram. And if your point falls safely
21:34
inside the curve closer to the origin, the part will hold.
21:38
Exactly. But if your point falls outside
21:40
the curve, the part fails. Fads beautifully bridged the gap
21:44
between pure fracture mechanics and pure yield mechanics.
21:47
I want to circle back to something we touched on earlier
21:49
with low cycle fatigue. There is a fascinating concept
21:52
called the transition life. Yes.
21:55
Imagine I'm designing a solid rocket motor casing for a
21:57
missile. It only has to survive a few 100
22:00
massive thermal and pressure cycles before its mission is
22:03
over and it's discarded. Right.
22:05
A very short, extreme life. Contrast that with a steam
22:08
turbine and a power plant that has to spin at 60 Hertz for 40
22:12
years. How does that vastly different
22:15
required lifespan change my material choice?
22:18
That specific dilemma is solved by the Manson Kaufman
22:21
relationship. This equation mathematically
22:24
relates the plastic strain amplitude experience during a
22:27
cycle to the total fatigue life of the part.
22:29
OK, so if you plot the elastic strain life and the plastic
22:32
strain life on a log log graph, they form 2 intersecting lines.
22:37
Right. And the transition life is the
22:39
exact intersection point, the number of cycles where the
22:42
damage from elastic plastic strain and the damage from
22:44
plastic strain are perfectly equal.
22:46
So how does knowing that transition life actually guide
22:49
the engineer's hand when picking a metal?
22:52
It dictates your primary material property requirement.
22:56
If your design life is vastly shorter than this transition
22:59
life, like your rocket motor casing, the plastic strain is
23:03
doing all the damage. It totally dominates the failure
23:06
mode. So what do you want in that
23:08
scenario? In that scenario, you actually
23:11
want a material with exceptionally high ductility.
23:14
You want a metal that can stretch, yield and absorb that
23:17
gross plastic deformation cycle after cycle without snapping.
23:21
But if your design life is much longer than the transition life,
23:24
like the steam turbine elastic strain dominates, the part is
23:28
hardly yielding at all. Exactly for that, ductility is
23:31
less important, and you need a material with an exceptionally
23:34
high ultimate tensile strength. It is a classic engineering
23:38
trade off them. It is highly ductile materials
23:40
are your best defense for short brutal design lives, while ultra
23:44
high strength materials are required for long repetitive
23:47
design lives. Which perfectly transitions us
23:49
into our next major topic. We spent all this time talking
23:52
about fatigue, which is entirely driven by cyclic mechanical
23:55
loads. Right, things moving back and
23:57
forth. But what about the slow, silent
24:00
enemies? The enemies of time,
24:03
temperature, and environment. What happens when the physical
24:07
operating environment itself turns hostile?
24:09
Oh, when the environment turns hostile, we step into the realm
24:13
of creep, fretting and corrosion.
24:15
The silent destroyers. Truly, these are the silent
24:18
destroyers because they don't require sudden impacts or
24:21
massive loads to ruin a design. Let's start with creep.
24:25
OK, creep, what is it? Engineers define creep as the
24:28
time dependent permanent plastic deformation of a material
24:31
subjected to a constant load at elevated temperatures.
24:34
You know this one bends the mind a bit.
24:36
If I hang a heavy 1000 LB weight on a steel rod in my garage at
24:40
room temperature, it stretches elastically by a fraction of a
24:43
millimeter and stops. And it will hang there for 1000
24:46
years without stretching another Micron.
24:48
Right, but if I introduce heat. Heat changes all the rules of
24:51
atomic bonding. If you heat that steel rod up to
24:54
roughly 35 to 70% of its absolute melting temperature, it
24:58
will continue to stretch. Really just from the heat.
25:01
Yes, very slowly, over thousands of hours, it will elongate.
25:07
Even if that 1000 LB load never changes, the thermal energy
25:11
allows the atoms in the crystal lattice to slowly diffuse,
25:14
sliding past obstacles that would normally stop them at room
25:17
temperature. So it's literally flowing like
25:19
an incredibly thick, slow moving liquid.
25:22
Exactly. And this flow, this creep
25:24
happens in three distinct metallurgical stages.
25:26
Let's go through them. First is primary or transient
25:29
creep. When the heat is first applied,
25:32
the deformation rate starts fast, but it gradually slows
25:35
down because the metal is strain hardening as it stretches.
25:38
So the dislocations in the crystal lattice this get tangled
25:40
up. Then it enters secondary or
25:43
steady state creep. The rate of stretching becomes
25:45
completely constant. The strain hardening is
25:48
perfectly balanced by thermal softening.
25:49
And the secondary stage is where the part spends the vast
25:52
majority of its useful operating life, right?
25:54
Yes it is, but eventually it hits tertiary creep.
25:57
The internal microstructure of the metal literally begins to
26:00
break down. Microscopic voids form at the
26:03
grain boundaries. Wow, so the remaining cross
26:05
section actually shings. Exactly.
26:07
The strain rate accelerates rapidly and the part inevitably
26:10
pulls apart in a catastrophic failure called stress rupture.
26:14
That is a terrifying reality for things like jet engine turbine
26:17
blades. I mean, they are spinning at
26:19
20,000 RPM under massive centrifugal loads, pulling them
26:23
outward, sitting directly inside a blazing inferno of combusting
26:28
jet fuel. It's the ultimate extreme
26:31
environment. They are constantly crooping,
26:33
growing microscopic fractions of an inch longer every flight
26:37
until they have to be retired before they strike the engine
26:40
casing. Yeah, creep is a massive design
26:42
factor there. But heat isn't the only silent
26:44
destroyer. What about fretting?
26:47
I think about a bearing press fit onto a shaft or two heavy
26:50
steel plates tightly bolted together.
26:52
Right, typical rigid joints. The joint is completely clamped
26:55
down. It isn't supposed to move at
26:57
all, but under dynamic machine loads it vibrates
27:00
microscopically. And those microscopic vibrations
27:03
are deadly. We are talking about minute
27:06
cyclic relative motions, often just a few thousandths of an
27:09
inch of slip. And that tiny rubbing causes
27:11
fretting action. Exactly.
27:13
Even machine surfaces look like jagged mountain ranges under a
27:17
microscope. When those two clamped surfaces
27:20
rubbed together, they micro Weld and tear off microscopic
27:23
asperities, the tiny piece of the metal surface.
27:26
And because it's metal tearing, you get debris.
27:29
But it's worse than just metal dust, right?
27:31
Much, much worse. Because those torn off
27:33
microscopic particles of fresh metal have massive surface area,
27:37
they immediately oxidize in the surrounding air.
27:39
So they turn into metal oxide like rust or aluminum oxide.
27:43
Right, which is essentially ceramic.
27:45
It is incredibly hard. This creates a hard, highly
27:48
abrasive debris that gets completely trapped in the tight
27:51
joint. So this debris acts like
27:52
grinding powder, accelerating the wear exponentially.
27:55
It does, but the wear isn't even the real danger.
27:58
The micro gouge is left behind on the surface of the metal act
28:01
as severe stress concentrators. Leading back to fatigue.
28:04
Exactly. This leads directly to fretting
28:07
fatigue, where a massive crack initiates from this hidden
28:10
pulverized vibrating joint and causes a catastrophic structural
28:14
failure at a load level vastly below the normal normal fatigue
28:17
limit. So it's mechanical wear creating
28:19
chemical oxidation, which then creates a highly abrasive powder
28:23
which causes more mechanical damage, which initiates A
28:26
fatigue crack. It's just a vicious cycle.
28:29
It really is. It feeds on itself.
28:30
Speaking of chemical damage, we have to talk about the entire
28:34
corrosion spectrum because a lot of people think corrosion just
28:37
means a rusty bumper on an old truck.
28:39
Rust is just the tip of the iceberg.
28:42
Direct chemical attack is the simplest, most uniform form of
28:46
corrosion. The metal just uniformly
28:49
dissolves, thinning out as it sits in an acid or aggressive
28:53
fluid. But as an engineer, the
28:55
insidious hidden forms of corrosion are much more
28:57
dangerous because they are localized.
28:59
Take crevice corrosion for example.
29:01
Oh, crevice corrosion is awful. It happens in tight, stagnant
29:05
spaces, like the gap under a rivet head, beneath a rubber
29:07
gasket, or inside a threaded hole.
29:09
Right, and the fluid inside that tiny gap gets starved of oxygen
29:13
compared to the surrounding fluid.
29:15
Yes, and this sets up a localized concentration cell.
29:18
The trapped fluid turns highly acidic and aggressively eats
29:20
into the metal hidden inside the crevice, while the surface out
29:24
in the open looks perfectly fine.
29:26
And then there is pitting corrosion, which is the absolute
29:28
bane of pipelines and chemical tanks.
29:30
Pitting is terrifying because it is an extremely localized attack
29:35
that drills deep microscopic holes straight down into the
29:38
metal. It completely bypasses the
29:41
surface armor. A tank can look pristine on the
29:43
outside but be perforated by thousands of microscopic pits
29:47
that compromise its pressure vessel integrity.
29:49
Exactly, and we also have to account for galvanic corrosion.
29:52
This occurs when 2 completely different metals are
29:55
electrically connected while sitting in an electrolyte like
29:58
seawater or even just salty Rd. slush.
30:01
So like a battery cell? Precisely.
30:04
The more chemically active metal acts as an anode and will
30:07
rapidly sacrificially dissolve itself to protect the more noble
30:11
cathode metal. Which is exactly why you don't
30:13
use stainless steel bolts to hold an aluminum boat hull
30:16
together in the ocean. The massive aluminum hull will
30:19
literally dissolve into a white powder around the bolt holes
30:22
trying to chemically protect the tiny steel bolt.
30:25
You see that happen all the time when people mix metals
30:27
improperly. But the real nightmare scenario
30:29
for an engineer isn't just one of these failing, it's when
30:32
these modes team up. What's fascinating here here is
30:34
the concept of synergistic failure.
30:37
When mechanical loads and chemical attacks combine, the
30:40
destructive result is far, far worse than the sum of its parts.
30:45
Give us an example. Take corrosion fatigue if you
30:49
have a part experiencing cyclic fatigue loads in a vacuum.
30:53
It lasts a certain amount of time.
30:55
If you put it in salt water with no load, it corrodes slowly.
30:58
OK, so isolated they are predictable.
31:01
Right, but if you apply fatigue loads while it's in a corrosive
31:04
environment, it will fail incredibly fast.
31:06
Why does it happen so much faster?
31:08
Because of the protective oxide layer, most metals protect
31:11
themselves by instantly forming a microscopically thin layer of
31:15
oxide that seals the surface. Like a natural shield.
31:17
Exactly. But under cyclic fatigue, the
31:20
metal strains and stretches. That brutal oxide layer cracks
31:24
during fresh, unprotected, highly reactive metal to the
31:27
corrosive fluid. And then the fluid attacks that
31:29
tiny spot. Right.
31:31
The fluid aggressively eats a microscopic pit into that fresh
31:35
metal. That new pit acts as a stress
31:38
concentrator, which accelerates the mechanical fatigue crack,
31:41
which stretches the metal further, breaking more oxide
31:44
bearing more metal. It's a feedback loop, and the
31:47
exact same mechanism happens in corrosion wear, right?
31:49
Yes, mechanical sliding. Physically, he scrapes off the
31:52
protective oxide layer, accelerating the chemical attack
31:56
exponentially. The chemical and mechanical
31:58
forces act as a feedback loop of destruction.
32:01
OK, so we understand the physics of twisting and bending.
32:04
We understand how fatigue cracks grow.
32:07
We know exactly how the environment, the temperatures
32:09
and the corrosive fluids are going to mercilessly attack our
32:12
design. We know what we're up against.
32:14
Armed with all that knowledge, how do we choose our armor?
32:17
When do I look at my CAD screen and say this bracket needs to be
32:20
aluminum or this linkage must be titanium?
32:22
Or break out the nickel super alloys for this turbine.
32:25
That decision process is the absolute crux of material
32:29
selection. Let's look at aluminum first,
32:31
because it is the undisputed workhorse of lightweight
32:33
structural design. How are the different types
32:36
organized? The aluminum alloy designation
32:39
system uses a four digit number from the one thousands to the 9
32:43
thousands to categorize them by their primary alloying elements.
32:47
So for example, the 2000s use copper, the five thousands use
32:50
magnesium, and the seven thousands use zinc for aerospace
32:54
grade strength. Exactly.
32:56
Then you upend the tempers. O means fully annealed and dead
33:00
soft, H means it was strain hardened by cold working, and T
33:05
means it was thermally treated or age hardened to lock the
33:09
crystal structure in a high strength state.
33:10
And aluminum's real superpowers are its manufacturing agility,
33:14
it's incredibly easy to machine, extrude and form, and its
33:17
strength to weight ratio. It's very easy to work with.
33:20
But the physical property that always blows my mind is it's
33:22
cryogenic performance. Most metals, particularly
33:26
standard body centered cubic carbon steel, have a ductile to
33:30
brittle transition temperature. Right, they get brittle in the
33:32
cold. You take standard steel, freeze
33:35
it in liquid nitrogen, hit it with a hammer and it doesn't
33:38
dent. It shatters like a glass window,
33:41
which historically is exactly what happened to the low grade
33:44
steel hull of the Titanic in the freezing Atlantic waters.
33:48
Yes, a classic example of low temperature embrittlement.
33:51
But aluminum, it actually gets tougher and stronger and extreme
33:55
cold. It does, and that is entirely
33:57
due to its crystalline architecture.
33:59
Aluminum has a face centered cubic crystal structure.
34:03
Because of how closely packed those atoms are, it does not
34:06
suffer from cold embrittlement. So it retains its ductility all
34:09
the way down to absolute 0. Exactly that makes aluminum the
34:13
undisputed king for cryogenic engineering.
34:16
It is the material of choice for the massive fuel tanks on the
34:18
space shuttle and modern rockets holding liquid oxygen and liquid
34:22
hydrogen at hundreds of degrees below 0.
34:24
But aluminum has an Achilles heel.
34:26
When you need high strength at elevated temperatures, it
34:29
completely fails you. It starts losing its structural
34:31
strength rapidly as you cross 200°C.
34:34
Yeah, it gets too soft. You can't use it near an engine
34:37
exhaust or in high speed aerodynamic friction.
34:40
That is where titanium's aerospace dominance begins.
34:43
Titanium is honestly a miracle metal.
34:46
It is roughly 40% lighter than steel, but can be alloyed to be
34:50
just as strong. It is incredibly corrosion
34:53
resistant. But to use it correctly, you
34:54
really have to understand it's crystallographic phases, right?
34:57
Yes, at room temperature, pure titanium exists in the alpha
35:00
phase, which has a hexagonal close pack crystal structure.
35:04
But if you heat it up past roughly 880°C, the crystal
35:09
lattice completely reshuffles itself into the beta phase.
35:11
Which is body centered cubic. Right, and by adding specific
35:15
alloying elements, metallurgists can stabilize these phases at
35:18
room temperature, creating alpha alloys, beta alloys, and the
35:22
highly desirable alpha beta mixed structures.
35:24
And the absolute undisputed king of the titanium world is
35:27
TI6AL4V, right? 6% aluminum, 4% vanadium.
35:32
It's an alpha beta alloy. It is the apex predator of
35:34
alloys. It accounts for more than half
35:36
of all titanium used globally. Why is it so popular?
35:39
The aluminum acts to stabilize the alpha phase while the
35:42
vanadium stabilizes the beta phase.
35:45
This creates A microscopic interlocking brick like
35:48
structure that gives it phenomenal high strength,
35:51
excellent fatigue resistance and allows it to be heat treated for
35:54
even more performance. It forms the backbone of modern
35:58
military fighter jets, commercial aircraft landing gear
36:01
and high performance medical implants.
36:03
It's everywhere in high end aerospace.
36:04
So the obvious question is, why don't we just use titanium for
36:08
absolutely everything? If it's as strong as steel but
36:11
nearly half the weight and doesn't rust, why are our cars,
36:15
bridges and skyscrapers still made of steel?
36:17
The primary barriers are astronomical costs and
36:20
fabrication difficulties. Extracting titanium from its
36:23
naturally occurring rutile ore using the Kroll process is an
36:26
incredibly complex multi day energy intensive.
36:29
It requires. Chlorine gas and a vacuum
36:32
environment. Yep.
36:33
Furthermore, machining a block of titanium is a machinist's
36:36
worst nightmare mayor because it conducts heat oorly.
36:39
The extreme heat of cutting doesn't flow into the metal
36:42
chips, it stays right at the cutting tool.
36:44
O the tool overheats. Badly.
36:47
The titanium reacts chemically with the carbide, cutting tools
36:49
at high temperatures, galling, smearing and literally welding
36:53
itself to the drill bits, snapping them instantly.
36:56
Wow. And while it handles heat vastly
36:58
better than aluminum, titanium isn't invincible.
37:01
It hits a very strict metallurgical temperature limit,
37:04
around 550°C. Yes.
37:07
Above that threshold, it rapidly absorbs oxygen and nitrogen
37:10
directly from the surrounding air, dissolving them into its
37:13
crystal lattice. And what happens then?
37:15
It becomes hopelessly brittle, forming a hard alpha case that
37:18
completely ruins its structural integrity.
37:20
So when temperatures exceed 550°C, like deep inside the
37:25
combustion Chamber of a gas turbine engine where the metal
37:28
is literally glowing white hot, what do we do?
37:31
We can't use aluminum. We can't use titanium.
37:33
Steel would just melt. That is when we call in the
37:36
heavy artillery, the nickel super alloys, materials like
37:40
Inconel 600, Waspoloy and the Monal family.
37:43
The really exotic stuff. Oh yeah, these alloys are
37:46
specifically almost magically engineered to maintain massive
37:50
tensile strength and resist creep deformation at
37:53
temperatures where standard steel would be a soft, glowing,
37:56
useless puddle. How do they survive that inferno
37:59
without creeping apart? They achieved this through
38:01
complex precipitation hardening. They are alloyed heavily with
38:05
elements like aluminum and titanium to form microscopic,
38:08
perfectly ordered intermetallic precipitates known as gamma
38:12
prime phases. Gamma prime phases, What do
38:15
those do? They act like millions of tiny
38:17
concrete roadblocks inside the nickel crystal lattice,
38:20
physically stopping the dislocations from sliding under
38:22
high heat. That's incredible.
38:24
Furthermore, by heavily alloying the nickel with chromium and
38:27
molybdenum, they form incredibly tenacious self healing oxide
38:31
layers that completely prevent high temperature oxidation.
38:34
And sulfidation too, right? Yes, they resist sulfidation,
38:37
which is the highly corrosive, aggressive chemical attack from
38:40
the sulfur contaminants found in jet fuel.
38:42
You know, I like to think of the metal families this way.
38:45
Aluminum is the lightweight sprinter, perfect for quick,
38:48
agile movements, mass production and surviving freezing
38:51
conditions. That works.
38:53
Titanium is the elite decathlete.
38:55
Incredibly strong, versatile and light, but very expensive,
38:58
offensive to train, and difficult to work with.
39:01
And Nickel is the heavily armored firefighter waiting
39:04
straight into the roaring flames, taking the absolute
39:08
worst thermal and chemical punishment imaginable without
39:11
flinching. That is a highly accurate and
39:13
vivid analogy. It perfectly summarizes the
39:17
metallurgical hierarchy, but it raises a critical design
39:20
threshold, which is what happens when we push those metals to
39:23
their absolute chemical and crystallographic limits to go
39:27
even lighter and stiffer than the best aerospace titanium.
39:30
We can't just hunt for a new metal on the periodic table.
39:33
We have to fundamentally rethink how matter is woven together.
39:36
Which brings us to a massive shift in engineering design.
39:39
Are synthetic fibers the future? What makes composites so
39:42
fundamentally different from the traditional metals?
39:44
We've spent the last hour discussing the.
39:46
Single defining characteristic that separates composites from
39:50
metals is anisotropy. Anisotropy.
39:53
Yes, the metals we've discussed, steel, aluminum, titanium, are
39:58
generally isotropic. That means they're macroscopic.
40:01
Material properties like stiffness, yield strength and
40:04
thermal expansion are the exact same in every single direction.
40:09
So if you pull the block of steel left to right, it resists
40:13
with the same strength as if you pulled it up and down.
40:16
Exactly. Composites, however, are
40:18
fundamentally anisotropic. Like natural wood, their
40:20
properties vary vastly depending on the direction of measurement.
40:23
Right, a carbon fiber composite is incredibly strong and stiff
40:27
along the longitudinal axis of the fibers, but it is remarkably
40:31
weak across them, perpendicular to the fibers.
40:33
Which completely changes the designers job with metal.
40:36
I draw the shape of the bracket and the metal does the rest.
40:39
But with composites, the engineer isn't just designing
40:41
the exterior shape of the part. You have to design the internal
40:45
architecture of the material itself at the exact same time
40:48
you design the geometry. Yes, you have to lay up the
40:51
plies of carbon fiber in precise orientations, zero degrees, 40-5
40:55
degrees, 90° to perfectly match the invisible load paths flowing
41:00
through the part. So what does this all mean for
41:02
how they fail? We spent a massive amount of
41:05
time talking about how metals fail in fatigue cracking from
41:08
cyclic tension. Do composites act the same way?
41:11
They behave almost entirely differently, which catches many
41:14
traditional metal engineers completely off guard.
41:17
While metals generally fail in tension.
41:19
Tension fatigue, where repeated pulling causes a microscopic
41:23
flaw to propagate into a fatal crack, composites are often far
41:28
more susceptible to compression. Compression fatigue.
41:30
Wait. Really.
41:31
Why compression? I would think pulling on the
41:33
fibers would snap them. It comes down to the micro
41:36
mechanics of the two distinct materials working together in a
41:39
composite. The incredibly strong carbon or
41:42
glass fibers provide all the structural strength and
41:45
stiffness. Right, the fibers do the heavy
41:46
lifting. The polymer matrix, the epoxy
41:48
resin, is relatively weak. It's only job is to bind the
41:51
fibers together, transfer the load between them and hold them
41:54
in exact alignment. OK, So what happens under
41:56
compression? Under secret compression, those
41:59
incredibly thin, stiff fibers want to buckle.
42:04
Imagine pushing down on the ends of a piece of dry spaghetti.
42:08
The epoxy matrix tries to keep them perfectly straight, but
42:11
over thousands of compressive cycles the matrix begins to
42:14
suffer micro cracking or the bond between the fiber and the
42:17
epoxy delaminates. When once the matrix loses its
42:20
firm grip. The fibers kink and buckle
42:22
inward, and the composite suffers a massive sudden
42:25
localized structural collapse. That is fascinating.
42:28
It's a completely different failure paradigm.
42:31
It's an instability failure at the microscopic level rather
42:34
than a cohesive bond failure. Exactly.
42:37
But it's not just strength and weight that make composites
42:39
valuable. They allow us to perform
42:41
absolute party tricks with Thermal Expansion, right?
42:44
Tailoring the CTE, the coefficient of thermal
42:47
expansion. Yes, and this is a total
42:48
superpower of composites. Most materials expand when
42:51
heated, but carbon fibers actually have a slightly
42:54
negative CTE along their axis, meaning they actually shrink
42:58
slightly when they get hot. Shrinking when heated is so
43:00
counterintuitive. It is the epoxy matrix, however
43:03
has a high positive CTE. It expands aggressively.
43:08
By carefully mathematically adjusting the volumetric ratio
43:11
of fiber to resin and the intricate angles of the fiber
43:14
layers, an engineer can create a composite part that has a
43:18
macroscopic CTE of exactly 0. It will not change size by a
43:22
single Micron, no matter how hot or cold it gets in operation.
43:26
Exactly, which is absolutely revolutionary and feels like
43:28
electronic packaging or space telescopes.
43:30
Like in high power electronics where the silicon processor
43:33
chips get incredibly hot, if the circuit board substrate they are
43:37
permanently mounted to expands at a different rate than the
43:40
silicon chip itself, that differential expansion will
43:43
literally sheer and snap the delicate microscopic solder
43:47
joints connecting them. Yes, by tailoring the CTE of a
43:51
composite substrate to perfectly match the thermal expansion of
43:54
the silicon, you eliminate those thermal stresses entirely.
43:57
And we aren't just limited to the standard carbon fiber and
44:00
epoxy pairings. We always hear about the polymer
44:02
matrix composites or PMCS. No.
44:04
Advanced engineering utilizes metal matrix composites or MMCS.
44:09
You can take a block of aerospace aluminum and
44:11
intimately infuse it with microscopic silicon carbide
44:15
ceramic particles. So combining a metal and a
44:17
ceramic. Exactly.
44:18
It dramatically boosts the specific stiffness and high
44:21
temperature wear resistance of the aluminum, essentially giving
44:24
you the ductile toughness of a metal combined with the
44:28
unyielding rigidity of a ceramic.
44:30
It's the ultimate blending of material families to defeat
44:33
specific failure modes. OK, so let's pull all of this
44:36
together. We understand the physical
44:38
forces of torsion and bending. We understand the failure modes
44:41
of fatigue, creep and corrosion. We have a vast pallet of
44:45
materials from cold loving aluminum to heat defying nickel
44:49
to anisotropic composites. The toolkit is huge.
44:52
The final daunting step of the engineering journey is
44:55
assembling all these millions of variables into the ultimate
44:58
optimized design. We can't just build a prototype,
45:00
break it, add more metal, and try again.
45:02
We can't rely on blind trial and error anymore.
45:05
Exactly. The stakes are too high and the
45:07
costs are too massive. We use mathematical design
45:10
optimization. This is where we hand the
45:12
physics over to the algorithms. And an optimization model
45:15
translates the physical world into pure math.
45:17
It does, and it has three core components.
45:20
First, the performance criterion, often called the
45:22
objective function. This is the one single
45:25
quantitative thing you are trying to strictly maximize or
45:28
minimal, like minimizing the total weight of an airplane wing
45:32
or maximizing the top speed of a race car.
45:34
What's the second component? 2nd the independent variables.
45:37
These are the specific dimensions, thicknesses or
45:40
material choices that the engineer or the computer has the
45:44
power to change during the search.
45:45
And the 3rd, 3rd and most importantly the system model
45:49
which consists of the constraints.
45:51
These are the absolute physics limits that cannot be violated,
45:54
like ensuring the maximum calculated von Mises stress
45:58
never exceeds the yield strength, or ensuring the part
46:00
geometrically fits inside a specific physical envelope.
46:04
But engineering in the real world is never that simple,
46:06
right? The objective function is never
46:08
just one thing. The boss always walks in and
46:10
wants it later. Andy cheaper, Andy stronger.
46:13
Andy manufactured faster. If we connect this to the bigger
46:16
picture, it becomes mathematically impossible to
46:19
optimize for everything simultaneously.
46:22
You cannot simultaneously minimize material cost, maximize
46:25
structural reliability, and minimize weight.
46:28
They conflict. Fundamentally, mathematically
46:31
conflict with one another. Improving 1 degrades the other.
46:34
To solve this, engineers use a sophisticated technique called
46:38
goal programming. Goal programming.
46:40
How does that work? Instead of strict optimization
46:43
of a single variable, you assign preemptive weights and priority
46:46
factors to your conflicting goals.
46:48
You are effectively telling the algorithm.
46:50
Goal one, avoiding catastrophic failure is critical, Goal 2,
46:54
minimizing weight is highly desirable, and goal 3, using a
46:57
cheaper alloy is just a bonus. Let me give a very practical
47:00
real world manufacturing example of this conflicting
47:03
optimization. Think about AC and C lathe in a
47:06
machine shop cutting a massive titanium part.
47:08
OK, a common scenario. The shop manager wants to
47:11
maximize the metal removal rate. They want to spin the part as
47:15
fast as possible and cut as deep as possible to save time and
47:18
increase throughput. Makes perfect sense for
47:20
business. But they also want to maximize
47:22
tool life, because replacing shattered carbide drill bits
47:25
costs a fortune and holds production.
47:28
But faster cutting inherently destroys tools faster.
47:32
How does the optimization algorithm mathematically handle
47:35
a conflict like that? The mathematical model uses what
47:38
we call deviational variables. It sets an idealized perfect
47:42
target for both the metal removal rate and the tool life.
47:45
An impossible target, basically. Exactly.
47:48
Then, during the simulation, it measures how far the actual
47:50
calculated performance deviates from those perfect impossible
47:54
targets. The algorithm applies penalties
47:57
for the underachievement of these specific goals, scaled by
48:00
the priority weights you assigned.
48:01
So it's trying to balance the penalties.
48:03
Right. The math navigates the complex
48:06
trade off curve, calculating millions of permutations until
48:09
it finds the precise optimal cutting speed and feed rate that
48:13
minimizes the combined penalties.
48:15
And because these engineering equations for 3D stress, fluid
48:19
dynamics, and thermodynamics are highly nonlinear, they curve and
48:23
exponentialize. Like the Paris law we discussed,
48:26
we have to rely on advanced computer algorithms to do the
48:29
heavy lifting. We can't solve these with a
48:31
pencil and paper. We absolutely cannot.
48:33
We rely on solvers using algorithms like Sequential
48:36
Quadratic programming or FQP. These algorithms are
48:40
specifically designed to handle highly nonlinear constraints and
48:43
objective functions. SQP That sounds intense.
48:46
Imagine the entire design space, every possible combination of
48:50
lengths, thicknesses and materials, as a massive multi
48:54
dimensional mountain range shrouded in thick fog.
48:57
OK, I'm picturing it. Every coordinate on the map is a
48:59
different design iteration. The elevation of the mountain
49:02
represents the performance of the part.
49:04
The SQP algorithm acts like a blindfolded climber dropped into
49:07
this range. So it can't see the peak at all.
49:09
No, but it feels the local slope of the terrain under its feet
49:13
using mathematical gradients, and it takes meticulously
49:16
calculated steps uphill, continually reassessing the
49:19
slope to climb to the exact topological peak of performance,
49:23
all while ensuring it never steps off the Cliff of a
49:26
constrained violation. It's incredible to think about
49:29
the journey we've just taken. We've gone from conceptualizing
49:32
the simple twisting shear of a mathematical shaft, to
49:36
understanding the microscopic growth of fatigue cracks, to
49:39
navigating an N dimensional topographical map with a
49:42
blindfolded mathematical climber just to find the perfect blend
49:46
of alloys and geometry. There's a lot to take in, but it
49:49
really highlights how deeply the science of mechanical design has
49:52
evolved. It is a constant, relentless
49:55
evolution. We are pushing the absolute
49:57
limits of known physics, chemistry and mathematics with
50:00
every single new generation of machinery, from deep sea
50:03
submersibles to orbital spacecraft.
50:05
We really are. We're asking more of material
50:07
than ever before. And that leads me to a final
50:10
provocative thought for everyone listening today.
50:13
We've spent this entire deep dive talking about optimizing
50:16
static shapes, picking the perfect rigid metal, or weaving
50:20
the strongest composite fiber. But what if the next massive
50:24
frontier of engineering isn't static at all?
50:26
I like where you're going with this.
50:28
Imagine a future of smart living composites, materials
50:32
deliberately embedded with microscopic sensor networks and
50:35
shape memory alloys that can dynamically alter their own
50:38
anisotropic properties in real time.
50:40
Like an organic system? Exactly.
50:43
Imagine a commercial airplane wing that senses a violent
50:46
turbulent gust, instantly shifts its internal matrix stiffness to
50:49
redistribute the stress away from a weak point, and then
50:52
after the flight artificially heals its own fatigue micro
50:56
cracks by releasing localized curing agents from embedded
50:59
micro capsules before the Paris can ever take hold.
51:02
That is exactly the horizon the most advanced materials labs are
51:06
running toward right now. It's a total blend of biology
51:08
and mechanical engineering, taking the concept of survival
51:11
and optimization down to a cellular level.
51:14
A transition from static dead structures to truly adaptive
51:18
living machines that respond to their environment just like an
51:21
organic body. And building that future all
51:24
starts with understanding the basic invisible forces we talked
51:27
about today. If you've got something useful
51:29
out of today's episode, follow and drop a like.
51:32
It helps grow the show and keeps the engineering knowledge
51:34
flowing.