Latest / Mechanical Engineering Made Simple / Why holes triple structural stress

Transcript

Why holes triple structural stress · Mechanical Engineering Made Simple

0:00
Welcome back to Sign Junkies to Mechanical Engineering Made
0:02
Simple. If the podcast brings you value
0:04
help, support the show with a like a follow or leave a comment
0:07
below. Glad to be here for another deep
0:09
dive. Yeah, so imagine a continent of
0:12
solid 2 inch thick steel, just a massive, perfectly uniform,
0:17
expansive metal, you know, stretching for miles.
0:20
Right, just infinite flat steel. Exactly.
0:23
Now imagine taking a microscopic like a needle fives drill bit
0:27
and poking a single invisible pinhole dead in the center of
0:30
that continent. Just a tiny void.
0:32
Yeah, if you put that steel continent under tension, the
0:35
physical stress at the exact edge of that microscopic barely
0:39
their whole doesn't just, you know, increase slightly.
0:41
Doesn't just bump up a bit. No, it immediately violently
0:44
triples. A tiny speck of nothingness
0:46
literally multiplies the forces tearing at the steel by a factor
0:49
of 3. It's wild to think about.
0:51
It is like, why? So today we are tearing down the
0:55
engineering physics of empty space.
0:58
We are diving into the mathematics of holes, stress,
1:01
concentration factors, and, well, the invisible battle
1:05
happening inside every piece of metal you've ever drilled into.
1:08
It really is. It's one of the most profound
1:10
paradoxes in mechanical design, honestly.
1:13
Yeah, because you take a perfectly good load bearing
1:16
material, right? And you intentionally remove a
1:18
piece of it, you create a void which.
1:20
Feels like you're just making it weaker.
1:22
Exactly. But that absence of material, it
1:25
completely redefines the structural reality of the entire
1:28
part. It's not just about the metal
1:30
you took away, it's about the geometric shockwave that the
1:33
void sends through the material that remains.
1:35
Right, because structural members with transverse holes
1:38
are just everywhere. Like if you were a mechanical
1:40
engineer, you just can't escape them.
1:41
You really can't. You're designing crankshafts
1:43
with oil galleries, leaf springs with clamping bolts, Ridge
1:49
gussets packed with rivets, aerospace skin panels riddled
1:52
with fastener, whole. They are ubiquitous in
1:54
engineering. So our mission for this deep
1:56
dive is to extract the absolute truth of how these empty spaces
2:01
disrupt stress. Because there's a lot of
2:03
confusion out there. So much we are getting answer
2:06
the questions that constantly plague engineering forums,
2:09
translate the intimidating charts into physical intuition,
2:12
and give you the factual reality of how to design structures that
2:15
actually survive the real world. And to build that intuition, you
2:19
know, we really have to reframe what a hole actually is.
2:22
Yeah, how so? Well, the fundamental trap I see
2:25
so many young designers fall into is viewing a hole merely as
2:29
a loss of cross-sectional area. Like just doing basic math on
2:33
the surface area. Right.
2:34
They think you know I removed 10% of the metal so my part is
2:37
10% weaker. And that's not true.
2:39
That is a highly dangerous underestimation.
2:42
A hole is a geometric obstacle. It forces the internal forces
2:46
the load paths running through the solid to violently detour.
2:50
OK, so if I'm visualizing this, I always think of it like a
2:54
crystal lattice. That's a good starting point.
2:56
If I'm pulling on a plate of steel, every iron atom is
2:59
holding hands with its neighbors, sharing that
3:02
microscopic load evenly across the entire width, right?
3:05
Exactly a nice even distribution.
3:07
But the moment I drill a hole, the atoms right at the boundary
3:11
suddenly have no neighbors on one side.
3:13
They're just grasping at air. Yeah, they reach out, grab
3:16
nothing, and so the entire physical burden they were
3:19
supposed to carry gets dumped instantly onto the atoms
3:22
directly adjacent to them, right on the edge.
3:25
That is a phenomenal way to conceptualize it, actually.
3:27
You're describing the disruption of the load path at the atomic
3:31
scale. It makes it feel more real.
3:33
It does because the stress isn't some abstract like magical aura,
3:37
it's literally the interatomic bonds stretching under excessive
3:41
burden right at the macro scale. The classic analogy is a wide,
3:44
rapidly flowing stream of water. OK, a stream.
3:48
If you drop a perfectly round, smooth boulder right into the
3:51
middle of that stream, the water doesn't just casually absorb the
3:55
obstacle. No, it has to rush around it.
3:57
Exactly. The water directly in front of
4:00
the boulder slows down to a halt, but the water squeezing
4:03
past the left and right sides of the boulder, it speeds up
4:07
immensely. It bunches together.
4:09
And that bunching is the stress. Yes, that violent bunching of
4:12
flow lines is exactly what we measure as the stress
4:15
concentration factor. We are calculating the ratio of
4:18
that peak localized bunched up stress to the baseline calm
4:24
water stress of the surrounding material.
4:26
OK, let's unpack this, because before we get deep into the
4:28
weeds of how different shapes change that bunching, we really
4:32
need to lay down our terminology for the class.
4:34
Definitely necessary. Because if you open any
4:36
engineering textbook or design manual, you are going to be hit
4:40
with specific notation. Specifically, the battle between
4:43
KPG and KTN. Yes, the two factors.
4:46
So KT obviously stands for the theoretical stress concentration
4:49
factor. But that G and that N they
4:52
change everything, right? They alter the entire
4:53
mathematical foundation of your safety analysis.
4:56
So the G stands for gross. Gross area, right?
4:59
KTG is the gross stress concentration factor.
5:03
This is the ratio you get when you compare the violent peak
5:06
stress right at the edge of the hole to the overall uncut gross
5:11
area of the part. Meaning you're looking at the
5:13
whole part. You are basically analyzing the
5:15
stress multiplier as if the hole wasn't even there stealing your
5:18
cross section. Which I mean, that feels a bit
5:21
like cheating the math. It is a bit of a theoretical
5:23
fantasy, yes. It's comparing the worst case
5:25
scenario to a fantasy world where you didn't even drill the
5:28
hole. So the alternative is KTM, the
5:30
net factor. Exactly.
5:32
I assume this means we are comparing the peak stress only
5:35
to the remaining solid material, like the actual minimum net
5:40
cross section that is physically leftover to carry the load.
5:42
You've got the distinction perfectly.
5:44
The net factor accepts reality. It deals with what's actually
5:47
there. Right.
5:48
It compares the peak stress to the nominal stress flowing
5:52
through the narrowed bottleneck of material beside the hole.
5:55
Makes total sense. And it might seem like you know
5:58
semantic bookkeeping right now, but later on when we discuss
6:02
materials undergoing cyclic fatigue, choosing the wrong
6:05
factor between gross and net won't just fail your exam, it
6:08
will cause your design to physically fracture in the
6:11
field. Yikes.
6:13
We will definitely circle back to that nightmare scenario, but
6:16
let's start at the absolute beginning, the baseline.
6:19
The classic starting point, the. Scenario every mechanical
6:22
engineering student stares at on a whiteboard during their junior
6:26
year. We have a single perfectly
6:29
circular hole in an infinitely wide, infinitesimally thin flat
6:34
panel. The Holy Grail of theoretical
6:36
stress analysis. And we are pulling that panel in
6:39
exactly One Direction. Pure uni.
6:41
Axial tension. Right.
6:42
To translate the math of the scenario into something
6:44
physical, we have to look at the elastic stress distribution
6:47
formulas derived in polar coordinates.
6:49
Sounds intimidating. It can be.
6:51
Now I won't torture everyone with the full differential
6:54
equations, but the physical take away is absolute.
6:56
OK, what is it? When you pull this infinite
6:59
plate, the stress distribution is not uniform around the
7:02
circle. The stress reaches its absolute
7:05
mathematical maximum at the exact edges of the hole that sit
7:08
perpendicular to the direction of your pole.
7:10
Let me lock that in. If I am gripping the top and
7:13
bottom of this steel plate and pulling it straight North and
7:15
South, the maximum stress isn't at the top and bottom of the
7:19
hole, is at the east and West edges the sides of the hole.
7:22
That is correct. Think back to the boulder in the
7:24
stream. Right, the water flowing.
7:26
The water isn't moving fastest at the front or back of the
7:29
rock, it's whipping around the extreme left and right flanks.
7:32
Oh, it's a great visual. And the mathematics dictate a
7:35
very specific undeniable ratio for those flanks.
7:38
The formula proved that the maximum stress at those east and
7:41
West points equals exactly 3 times the nominal applied
7:45
stress. Three times.
7:46
Yes. Therefore, your baseline
7:49
theoretical stress concentration factor, your KT, equals a
7:53
perfectly clean 3. Now here is where my internal
7:57
skeptic flares up, and it's the exact same pushback I see
8:00
engineers posting on forums all the time.
8:02
I think I know what you're going to ask.
8:03
Let's take the theory out of the textbook and put it into my
8:06
continent of steel. If I have a massive infinitely
8:11
wide plate of solid metal and I pull it and the stress is say
8:15
10,000 PSI everywhere. If I take a microscopic drill
8:19
bit a fraction of a millimeter wide and I plunge it into the
8:23
dead center of that steel continent.
8:25
A tiny, tiny pinhole. Yeah.
8:27
Does that tiny, almost invisible speck of missing metal really
8:31
triple the stress at its edge to 30,000 PSI?
8:35
Yes, it absolutely does. Seriously.
8:37
And that is the exact conceptual hurdle that trips up incredibly
8:41
smart people. The multiplier That factor of 3
8:44
is dictated entirely by the geometry of the obstacle, not by
8:47
its absolute physical size. The size doesn't matter at all
8:50
for the multiplier. A microscopic pinhole behaves
8:52
exactly the same way as a circular hole the size of a
8:55
football stadium, provided the surrounding panel is infinitely
8:58
wide relative to the hole. It is hard to wrap my head
9:00
around. Think of it this way, the water
9:03
rushing around a tiny grain of sand experiences the same
9:07
proportional geometric bunching as water rushing around a
9:10
massive boulder relative to the scale of the object.
9:13
It just feels wrong. It feels like a tiny pinhole
9:17
shouldn't have the leverage to triple the stress of a massive
9:20
plate. It feels wrong because our
9:21
brains naturally conflate the intensity of the stress with the
9:24
volume of material affected by it.
9:27
OK, and that brings us to the saving grace of mechanical
9:29
engineering. Stress decay.
9:31
Yes, the stress right at the boundary of that microscopic
9:34
pinhole triples to 30,000 PSI, but we have to analyze how far
9:40
that triple stress actually penetrates into the surrounding
9:42
steel continent. The data proves that this
9:45
multiplier is intensely localized.
9:47
Meaning the damage is contained. Atoms a few inches away have no
9:51
idea the hole even exists. Exactly the right framing.
9:54
Let's look at the mathematical decay of that circumferential
9:57
stress as you move away from the hole along that east West axis.
10:01
If your hole has a radius of a, then by the time you travel a
10:05
distance of just five times them from the center.
10:07
So 5 radiate out. Yes, by that point the
10:10
circumferential stress has plummeted.
10:12
It has decayed down to 1.02 times the nominal stress.
10:17
Oh wow, so the drop off is practically a Cliff.
10:20
It is very steep. If I have a one inch radius
10:22
hole, by the time I am 5 inches away from the center, the stress
10:25
multiplier is basically gone. It's just 2% higher than normal.
10:28
Exactly. And if you have a microscopic
10:31
pinhole with a radius of a thousandth of an inch, the
10:34
stress is back to normal, just five thousandths of an inch
10:38
away. That's tiny.
10:39
The boulder only causes violent Rapids immediately adjacent to
10:42
its surface. A few feet down the bank, the
10:45
water is flowing smoothly and evenly.
10:47
Again, that's a. Huge relief.
10:49
This violently localized nature of stress concentration is the
10:53
only reason human civilization can build things.
10:56
It's why you can drill a hole in a steel beam to hang a light
10:59
fixture without the entire skyscraper instantly shearing in
11:03
half. Because the damage doesn't reach
11:04
the edges of the beam. Exactly.
11:06
The elevated Truss is real, but it is confined to a tiny
11:10
microscopic Halo around the cut. That makes so much more sense.
11:15
The peak is high, but the mountain is incredibly narrow.
11:18
Precisely. But, you know, let's leave the
11:21
realm of theoretical continents of steel.
11:25
In reality, my steel plates have finite widths.
11:28
Yes, the real world. I'm usually designing like a
11:31
narrow metal strap, like a tension bracket where I have to
11:34
drill A1 inch bolt hole right through the middle of a 2 inch
11:39
wide piece of flat bar. So the hole is half the width.
11:42
Yeah, the hole is taking up half the real estate.
11:45
I can't imagine AKT of three applies there because the stress
11:49
doesn't even have 5 radii of room to decay before it hits the
11:52
outside air. It doesn't apply, and this is
11:55
where the math transitions from elegant clean polar coordinates
11:58
into the gritty reality of empirical testing.
12:01
Real world data. Yes.
12:03
When your whole starts consuming a significant fraction of your
12:06
panel width, the outer boundaries of the panel start
12:09
fighting back. Fighting back the stress flow
12:11
Lines are being squeezed between the edge of the hole and the
12:13
edge of the part. They can't flare out naturally
12:15
into an infinite plane. Oh, they are tracked in a
12:17
bottleneck. So if the factor isn't 3, how on
12:20
earth do we calculate it? This is where we rely on the
12:23
foundational work of researchers like RCJ Howland and later the
12:27
brilliant empirical formulas developed by RB Haywood Haywood.
12:31
Yes, Haywood ran the numbers and the tests to cover this finite
12:35
width reality. He developed an equation that
12:37
states the net stress concentration factor KTN equals
12:42
2 plus the cube of the quantity 1 minus the ratio of your hole
12:46
diameter to your panel width. OK, audio math is always tough.
12:50
Let's translate that into a physical tug of war.
12:52
Sure, we have our hole diameter, let's call it D.
12:55
We have our total panel width H Haywood is basically saying we
12:58
need to look at the ratio of D / H.
13:00
Yes, that ratio is the key. If I have a tiny hole in a wide
13:03
plate, D / H is practically 0. Correct.
13:07
And if you run the math, 1 -, 0 is 1, the cube of one is 1, and
13:12
2 + 1 brings us right back to our theoretical baseline of
13:15
three. Oh wow, so the formula holds up
13:17
perfectly for wide plates. It matches the infinite plate
13:20
theory. It provides a perfect bridge to
13:21
the infinite theory. But let's push it to the
13:24
extreme. What if my hole diameter D is
13:26
getting massive? Like it's chewing up the whole
13:29
bracket, the diameters approaching the total width H.
13:32
So mostly whole, very little metal, yeah.
13:35
That ratio of d / H approaches 1.
13:37
Let's follow the physics. If the ratio is 1, then 1 -, 1
13:41
is 0. The cubed term completely
13:44
vanishes. Which means the KTN approaches 2
13:48
way it goes down. It does.
13:49
As the hole gets bigger and eats more material, the stress
13:52
multiplier drops from three down to 2.
13:54
It does, based strictly on the net area.
13:57
The geometric multiplier softens because the material is acting
14:00
less like a plate with a hole and more like 2 parallel strips
14:04
of metal. Oh, I see.
14:05
But this mathematical quirk reveals a truly fascinating
14:08
physical phenomenon, one that was observed in deep detail by
14:12
researchers like AM Wall and RB Wooks.
14:14
What do they see happening to the metal?
14:16
Imagine your hole diameter is 99% the width of your panel.
14:20
What you have left on the left and right sides is not a flat
14:23
plate anymore. No, obviously.
14:24
It is just two microscopic hair thin filaments of metal trying
14:29
to hold the entire. Load just tiny little slivers of
14:33
steel spanning the gap. Yes, and Wall and Bukeys observe
14:36
that under load, particularly when dealing with finite
14:40
deformations in plastic models, something incredible happens.
14:43
What happens? Those infinitely thin filaments
14:45
don't just sit there and take the eccentric because they are
14:49
so thin, they physically warp. They bend inward toward the
14:53
center of the empty hole as the part stretches longitudinally.
14:57
Wait, they acting like strings being pulled taut?
15:00
When you pull a loose string, it straightens out to perfectly
15:02
align with the force. That is the exact mechanism.
15:05
Because those slivers physically move and realign their geometry
15:08
with the primary load vector, they eliminate the eccentric
15:12
bending forces. They allow the stress to
15:14
distribute almost perfectly uniformly across their tiny
15:18
microscopic cross sections. Meaning the stress concentration
15:21
multiplier drops even lower. Theoretically, yes.
15:24
For large plastic deformations where the material can flow and
15:27
realign, the KTN can theoretically plummet all the
15:30
way down to a perfect 1.0. 1.0. The multiplier vanishes entirely
15:36
because the geometry has essentially deformed into two
15:39
straight wires pulling in perfect tension.
15:42
That is a wild survival mechanism built into the physics
15:45
of the material. It is, but I must add a massive
15:48
blinking caution light for any engineer listening.
15:51
A warning label. A huge one.
15:52
This drop to 1.0 only happens in large plastic deformations,
15:58
meaning that the part is already permanently yielding and
16:00
stretching out of shape. It's already failing basically.
16:03
Yes, for tiny rigid elastic deformations, the safe zone
16:08
where we actually design parts to live the inward movement
16:12
isn't significant enough to relieve the bunching.
16:15
In the elastic zone. The KTN bottoms out and stays
16:18
stubbornly at 2. Got it.
16:20
Furthermore, as researchers believe and Apple corruptly
16:22
point out, if you have a hole that consumes 99% of your parts
16:26
with the microscopic drop in your theoretical stress factor
16:29
is entirely irrelevant. Because your part is going to
16:32
snap anyway if you leave a hair thin sliver of metal to hold up
16:36
a tractor. It doesn't matter if the
16:38
multipliers 2 or 1, the baseline stress is in the millions.
16:42
Precisely the point. The net area is so
16:45
catastrophically small that failure is guaranteed regardless
16:48
of the multiplier, but it serves as a beautiful demonstration of
16:52
how geometry and force constantly negotiate with each
16:55
other. I love that the material is
16:58
actively trying to find the most efficient load path.
17:00
It really is. But you know that localized
17:03
stress decay is fascinating for a flat plate in a tensile
17:07
testing machine. In the real world, my designs
17:11
aren't just being pulled in One Direction.
17:13
Very rarely are they. If I'm designing the pressure
17:15
hall of a deep sea submersible, the ocean isn't just pulling it
17:18
end to end, it's crushing it from all sides simultaneously.
17:22
True 3D loading. How does this boulder in a
17:24
stream analogy hold up when the stream is hitting you from every
17:27
possible direction? It completely overwrites the
17:30
stress map. When we introduce biaxial
17:32
stress, loading the material in two perpendicular directions
17:35
simultaneous, we enter a much more complex arena.
17:38
I can imagine to navigate it, engineers rely on the principle
17:41
of superposition because we are operating in the linear elastic
17:45
range of the metal, meaning it springs back to its original
17:48
shape when you let go. We don't have to invent entirely
17:52
new exotic math. So if I'm guessing how
17:55
superposition works, if I know the stress field for a
17:59
horizontal pull and I know the stress field for a vertical
18:02
pull, I could literally just lay those two mathematical maps on
18:06
top of each other and add the numbers together.
18:07
You've hit the nail on the head. Superposition is the
18:10
mathematical equivalent of stacking transparent maps.
18:13
Let's build the model. We pull horizontally on the
18:16
plate. With a primary stress we will
18:17
call Sigma one. We pull vertically with the
18:20
secondary stress we will call Sigma 2.
18:23
To make our equations universal for any load, we define a simple
18:26
ratio Lambda. Lambda is simply Sigma 2 divided
18:30
by Sigma 1. So Lambda is just our gauge for
18:32
how hard the vertical pull is fighting against the horizontal
18:35
pull. If the vertical pull is half as
18:37
strong, Lambda is 0.5. Exactly.
18:40
Now, using that stacked map of superposition, we can predict
18:44
the peak stress at any point on the whole's edge.
18:46
OK, let's hear the formula. At the top and bottom edges of
18:49
the circle, the peak stress is equal to Sigma 1 multiplied by
18:53
the quantity 3 times Lambda -1. Conversely, the peak stress at
18:58
the left and right edges of the circle equals Sigma 1 multiplied
19:02
by the quantity 3 minus Lambda. OK, I want to plug some real
19:06
visceral numbers into this because I know this leads to an
19:09
absolute aha moment for anyone designing pressure vessels.
19:14
It really does. Let's look at equal biaxial
19:17
tension. I am inflating a perfect
19:20
metallic sphere like a high pressure gas storage tank.
19:23
The internal pressure is pushing out equally in every single
19:26
direction. A.
19:27
Classic spherical pressure vessel.
19:29
That means the horizontal stress Sigma 1 is exactly equal to the
19:33
vertical stress Sigma 2. They are identical, which means
19:36
our ratio Lambda equals exactly 1.
19:38
So let's run your formula for the left and right edges.
19:40
It was Sigma one times the quantity 3 minus Lambda.
19:44
Yes, 3 -, 1 is 2, which means the maximum stress concentration
19:47
factor drops from a three down to A2.
19:49
The math holds and if you check the top and bottom edges using
19:53
the other formula three times Lambda -1.
19:56
Okay, 3 * 1 - 1. That also equals 2.
19:58
The stress factor is 2 everywhere.
20:01
It is perfectly uniform all the way around the entire
20:04
circumference of the hole. Wow.
20:06
Because you are pulling equally in all directions, the geometric
20:09
bunching effects are essentially interfering with each other.
20:13
The horizontal pole wants to bunch the stress at the top and
20:16
bottom, while the vertical pole wants to bunch the stress at the
20:19
left and right. And they just cancel out.
20:21
And they fight each other to a draw, and the overall stress
20:24
state settles into a highly stable uniform KT of two.
20:28
That is incredibly elegant, but what if the forces aren't
20:31
playing nice? They often don't.
20:33
What if I'm not inflating a sphere, but twisting a panel?
20:37
What if I'm stretching the metal left to right but simultaneously
20:40
squishing it? Compressing it top to bottom.
20:43
That is the notorious state of cure shear, and it is violent.
20:48
In this scenario, your vertical stress Sigma 2 is the exact
20:52
negative of your horizontal stress Sigma 11 is pulling, the
20:57
other is pushing with the exact same force.
20:59
Therefore our ratio Lambda equals -1 Oh boy, let.
21:04
Me. Run the mental math on that.
21:06
For the left and right edges 3 minus Lambda.
21:09
That becomes 3 - -1 correct, which is 3 + 1 four.
21:14
The KT jumps to 4. It jumps to four and it creates
21:17
an incredibly volatile situation.
21:20
The tension and compression fields are no longer fighting
21:23
each other to a draw. They are constructively
21:26
interfering. Making it worse.
21:27
Much worse, the geometric squeezing of the compression
21:30
adds to the geometric stretching of the tension, conspiring to
21:33
RIP the hole apart at the seams. Which is a brilliant proof of
21:36
why you can't just be lazy as an engineer.
21:39
Slap a standard three times safety factor on every hole in
21:42
your CAD model and call it a day.
21:44
You really can't. You have to.
21:45
Trace the load path. You have to know how the part is
21:47
being loaded, not just that it's being loaded.
21:48
Exactly right. But let's add another layer of
21:50
reality. We talked about spheres and flat
21:53
plates. What about cylinders?
21:55
Cylindrical shells. Most of the pressurized things
21:58
we build, rocket bodies, submarine holes, oil pipelines
22:02
are cylinders. Wrapping a flat plate into a
22:06
tube has to mess with the physics.
22:08
It introduces an entirely new highly destructive variable
22:13
curvature. Curvature.
22:14
A flat plate, even a biaxially loaded 1, is a highly idealized
22:18
scenario because all the forces act entirely within the flat
22:22
plane of the material. But the moment you wrap that
22:25
plate into a cylinder and pressurize it, the physics
22:27
change. Because it's trying to unroll.
22:29
You still have membrane stresses stretching the metal, but now
22:32
you introduce immense bending stresses right at the edge of
22:35
the hole. Because the hole is essentially
22:37
a weak spot that breaks the structural continuity of the
22:39
arch, it wants to bulge out radially away from the center of
22:43
the pipe. That is the exact physical
22:45
mechanism. The solid metal around the hole
22:47
has hoop strength to resist the internal pressure, but the empty
22:50
void does not. So it pushes out.
22:52
So the internal pressure pushes the flexible edges of the hole
22:55
outward, creating a local bulge. To calculate the severity of
22:59
this bulging stress, aerospace and pressure vessel engineers
23:03
rely on a critical dimensionless parameter called epsilon,
23:07
derived from shallow thin shell theory.
23:10
OK, hit me with epsilon. How do we weigh this bulge?
23:13
Think of epsilon not as a string of random numbers, but as a
23:16
mathematical scale balancing two different geometric realities.
23:21
Epsilon equals a constant 0.639 multiplied by the ratio of your
23:26
whole radius to the mean radius of your cylindrical.
23:29
Shell whole radius over shell radius.
23:31
Then you multiply all of that by the square root of a second
23:34
ratio, the thickness of your shell divided by its mean
23:37
radius. So it's heavily dependent on how
23:39
thick the wall of the pipe is compared to how big the pipe is
23:42
overall. Why is this specific messy
23:45
parameter the gold standard for an engineer designing a rocket?
23:49
Because Epsilon serves as the ultimate boundary guard, it
23:52
dictates the region of validity for all your assumptions.
23:55
Let's look at aerospace. If you are designing a rocket
23:58
body or a commercial airplane fuselage, you are dealing with
24:02
incredibly thin aluminum or composite shells over massive
24:05
diameters. The shell thickness is tiny
24:08
compared to the overall radius. Like a soda can.
24:11
Exactly. In a soda can scenario, this
24:14
geometry drives your epsilon value to relatively large
24:17
numbers, say 2-3 or four. And what does a large epsilon
24:21
mean? When epsilon is that high, the
24:23
theoretical KT factors in the engineering charts absolutely
24:26
skyrocket. They detach from the flat plate
24:28
baseline entirely. Oh wow.
24:30
Because the shell is so thin, it offers almost no bending
24:33
stiffness. The bulging effect at the hole
24:35
is severe. You aren't dealing with a gentle
24:37
flat plate KT of two or three anymore.
24:40
What are we dealing with? The bending stress might drive
24:42
your local multiplier to a factor of 5-6 or even higher.
24:45
That's terrifying, but what if I'm on the other end of the
24:48
spectrum? What if I'm not designing a thin
24:50
aerospace soda can, but a massive thick walled hydraulic
24:55
pipe for an excavator? In that heavy machinery regime,
24:58
your wall thickness is very large relative to your cylinder
25:02
radius. If you plug those chunky numbers
25:04
into the formula, epsilon plummets to a very small
25:08
decimal, like 0.5 or less. And when epsilon is tiny.
25:12
The caricature effectively ceases to matter.
25:15
Really. The thick walls possess so much
25:17
inherent bending stiffness that the hole cannot physically bulge
25:22
outward. The structural integrity holds
25:24
the curve, and the thick pipe behaves almost identically to
25:28
the theoretical flat plate we started with.
25:30
It's fascinating. Epilon is a beautiful metric
25:32
because it mathematically guides your physical intuition, telling
25:35
you exactly when curvature is a deadly threat and when it's
25:38
negligible. Here's where it gets really
25:40
interesting. Speaking of physical intuition,
25:42
let's follow the logic trail to a completely different shape.
25:45
Different from a circle. Yeah, because my intuition says
25:48
if a circular hole causes a baseline 3 times stress
25:51
multiplier, and even in the best case biaxial load it still
25:55
causes a two times multiplier, why on earth are we so obsessed
25:59
with drilling circles? It's a great question.
26:01
If circles are such terrible structural bottlenecks, why do
26:04
we use them everywhere? Which brings us to perhaps the
26:08
most famous engineering trivia question of all time.
26:11
Why are airplane windows Oval instead of round or square?
26:16
It is the ultimate story of engineering optimization forged
26:19
in tragedy, and it all revolves around the mathematical
26:22
properties of the ellipse. The ellipse, the theoretical
26:25
baseline for the elastic stress distribution around an
26:28
elliptical hole in an infinite plate, was cracked by the
26:31
British engineer Charles Inglis back in 1913.
26:35
The Inglis formula. I know this is foundational.
26:38
How did he map the stress for a stretched out circle?
26:41
Inglis realized that an ellipse gives us 2 levers to pull,
26:45
rather than just the single radius of a circle.
26:47
OK, 2. Levers.
26:48
Let's assume we have an elliptical hole with a
26:50
horizontal semi major axis we'll call A, and a vertical semi
26:53
minor axis we'll call B. So it's an Oval sitting flat on
26:57
its long side. Got the picture?
26:58
If we take our plate and pull it with a uniaxial tension straight
27:01
up and down, pulling it perpendicular to that long A
27:04
width, Inglis proved that the gross stress concentration
27:08
factor KTG equals 1 + 2 times the ratio of A / B.
27:13
OK 1 + 2 * A / B. Yes.
27:17
Alternatively, you can express it in a way that feels more
27:19
intuitive for machinists and designers, using the actual
27:23
sharpness of the curve at the edge.
27:25
How does that look? It's mathematically equivalent
27:27
to say KTG equals 1 + 2 times sqrt a / r where R is the radius
27:34
of curvature at the sharpest extreme of the ellipse.
27:37
I want to stress test this formula.
27:38
Let's put it through some extreme boundary conditions to
27:41
see if the physics hold. Up.
27:42
Let's do it. Let's start by morphing our
27:44
ellipse back into a perfect circle.
27:46
In a perfect circle, the width A and the height B are exactly
27:49
identical. They are.
27:50
So the ratio of a / b is simply one.
27:53
The formula becomes 1 + 2 * 1. Which is 3.
27:57
We arrive right back at the baseline multiplier for a
27:59
circular hole. Inglis's math is perfectly
28:02
backward compatible. It is.
28:04
Now let's take it to the other terrifying extreme.
28:08
Let's squeeze the ellipse. Let's make the height incredibly
28:11
tiny, approaching 0, while the width A remains large.
28:15
Visually, you have an infinitely narrow horizontal slit in the
28:20
metal. If I'm visualizing that it's no
28:22
longer a hole, you just described a razor sharp crack
28:26
propagating through the steel. I did and look at what the math
28:29
does when B approaches 0. The ratio of a / b involves
28:34
dividing a whole number by a near 0 decimal.
28:37
Which goes to Infinity. The resulting fraction explodes.
28:40
The stress multiplier skyrockets exponentially toward Infinity.
28:44
The shark of the crack, the more the stress multiplies, tearing
28:47
it even further open. Exactly.
28:48
This is the entire foundation of linear elastic fracture
28:51
mechanics. A sharp crack creates a near
28:54
infinite stress concentration at its microscopic tip.
28:57
That's why cracks are so deadly. The interatomic bonds at that
29:00
dip are experiencing forces hundreds of times higher than
29:02
the surrounding metal, which is why once a crack initiates, it
29:06
takes very little additional force to unzip the metal
29:08
completely. Which is exactly what happened
29:10
to the de Havilland Comet. We have to talk about that
29:12
history because it proves why we care about these shapes.
29:16
A very dark day for aviation. The Comet was the world's first
29:20
commercial jetliner in the 1950s.
29:22
It was a marvel. But then they started falling
29:26
out of the sky, exploding midair.
29:29
It remains one of the most sobering lessons in modern
29:32
engineering. The Comet was designed with
29:35
square passenger Windows 8. Square windows.
29:38
Now, they weren't perfectly sharp squares.
29:40
They had slightly rounded corners, but as we just
29:43
discussed, a tight radius of curvature acts identically to a
29:47
tight ellipse. That sharp R value in the
29:49
formula. The engineers at the time didn't
29:51
fully appreciate the magnitude of the stress bunching at those
29:54
square corners under cyclic pressurization.
29:57
Every time it took off. The airplane would take off, the
29:59
cabin would inflate like a balloon, the corners of the
30:02
square windows would experience massive stress concentrations,
30:05
and over hundreds of flights, microscopic fatigue cracks
30:08
initiated at those corners. Eventually, the fuselage
30:12
violently unzipped at altitude. So how did the industry fix it?
30:16
Walk us through the optimization of the modern airplane fuselage
30:19
using Inglis's insights. Why vertical ovals?
30:23
To solve the window problem, we have to treat the airplane's
30:26
fuselage for what it truly is, a massive flying pressure vessel.
30:31
OK. But pressure vessel.
30:32
It is a sealed cylinder maintaining A survivable
30:35
atmosphere while cruising through a low pressure void.
30:39
The internal cabin pressure is desperately trying to tear the
30:43
aluminum skin apart in two distinct directions
30:45
simultaneously. Biaxial stress again.
30:48
First you have the hoop stress. This is the tangential force
30:51
pushing radially outward trying to split the cylinder open down
30:54
the middle like. A hot dog bursting its casing on
30:56
a grill. A vivid, if messy, physical
30:58
analogy. Second, you have the axial
31:01
stress. This is the longitudinal force
31:04
acting on the pressure bulkheads at the nose and tail, trying to
31:07
stretch the entire airplane lengthwise like a rubber band.
31:10
Now we are back to our biaxial tension model from earlier.
31:13
But I'm guessing this is an equal biaxial tension.
31:15
Like our sphere, the forces aren't balanced.
31:18
They are drastically unbalanced due to the basic physics of a
31:22
pressurized cylinder. The hoop stress, the bursting
31:24
force is always exactly twice the magnitude of the axial
31:28
stretching force. Twice as much.
31:30
So using our superposition notation, we have a biaxial
31:33
state where Sigma 2, the axial stress equals Sigma one, the
31:36
hook stress divided by exactly 2.
31:39
The vertical pole stretching the plane is only half as strong as
31:42
the horizontal pole trying to burst it.
31:44
Precisely. Now, if you are a naive designer
31:46
and you punch a perfectly circular window into this
31:49
unbalanced stress field, the physics fight you.
31:52
The uneven forces create uneven bunching around the circle.
31:55
What's the number? If you run the mathematics, the
31:58
maximum localized stress concentration for that circular
32:01
window hits roughly 2.5 times the nominal stress. 2 1/2 times
32:06
That is a massive multiplier, and the only way an aerospace
32:09
engineer can survive a multiplier that high is to
32:11
thicken the entire fuselage. Add more aluminum, add more
32:15
weight. Which is the mortal enemy of
32:17
aircraft design? It is incredibly inefficient.
32:20
A circular window is carrying a multiplier that is 1.666 times
32:25
higher than it theoretically needs to be.
32:27
But this is where Inglis's mathematical lever comes in.
32:30
How do we fix it? We want the stress to be
32:32
perfectly uniform all the way around the edge of the window.
32:35
We want to eliminate the hotspots to achieve that balance
32:39
in a stress field where the horizontal force is twice the
32:41
vertical force, we must design A shape that perfectly mirrors
32:45
that imbalance. So if the horizontal stress is
32:47
twice as bad, we make the vertical height of the window
32:51
twice as tall to give the horizontal stress a smoother,
32:54
gentler path to flow around. That is the exact optimization
32:58
mechanism. We design an elliptical window
33:00
where the vertical height A is exactly twice the horizontal
33:04
width B. The ratio is 2 to 121.
33:06
When you align that vertical Oval with the massive hoop
33:09
stress, the stretched out geometry perfectly counteracts
33:12
the unbalanced forces. The tangential stress flowing
33:14
around the window smooths out entirely.
33:17
The theoretical KT drops to an incredibly efficient uniform 1.5
33:22
all the way around the perimeter.
33:23
It's just brilliant. It's geometry weaponized against
33:26
physics. It's the absolute optimum shape
33:29
and that is why you look out of a vertical Oval when you fly.
33:32
It's a beautiful solution. But let me throw a wrench in the
33:34
works. Manufacturing perfect
33:37
mathematical ellipses is, well, it's annoying.
33:39
Machinists aren't always fans. What if a machinist wants to use
33:42
a shape that's easier to cut, say a slot with straight sides
33:47
and semicircular ends, which I believe is termed an ovaloid or
33:51
a square opening with heavily rounded corners?
33:54
How do those non elliptical shapes behave?
33:56
This brings us to a highly practical concept in stress
33:58
analysis called the equivalent ellipse.
34:01
You can accurately predict the stress concentration of a
34:04
complex rounded geometry by mathematically matching it to an
34:08
imaginary perfect ellipse that tightly envelops its outer
34:11
dimensions. Wait, really?
34:13
You're saying a square with rounded corners behaves
34:15
identically to a smooth ellipse. To the flowing stress field,
34:19
yes, they are largely indistinguishable.
34:22
Consider the physical corners, the cusps of that rounded square
34:25
that protrudes slightly outside the boundary of our imaginary
34:28
enveloping ellipse. The extra bits of metal.
34:30
When researchers conduct photoelastic tests on these
34:33
complex shapes, where they machine the shape out of special
34:36
translucent plastic and pass polarized light through it while
34:40
under load, we can physically see the stress fringes light up
34:44
in vibrant colors. Like a heat map?
34:46
Exactly. And we see something remarkable
34:48
at those protruding corners. They become photoelastically
34:52
dead. Dead, meaning the plastic isn't
34:55
lighting up. Meaning they carry almost zero
34:57
stress. The stress flow lines rushing
34:59
around the massive obstacle take the smoothest, most direct path
35:04
of least resistance. I see they naturally trace an
35:07
elliptical path, completely bypassing the physical corners
35:10
of the square. The material sitting in those
35:12
corners is essentially structurally useless.
35:15
It is bypassed by the load, leaving it as a low stress dead
35:18
region. Therefore, the peak stress
35:21
concentration of the rounded square is practically identical
35:24
to the theoretical ellipse that encompasses it.
35:26
That is wild. The stress flow literally
35:28
ignores the material that isn't aerodynamic.
35:31
It carves its own efficient path.
35:33
Nature is lazy. It takes the easiest path.
35:35
OK, but a commercial airplane isn't just a tube with a few
35:38
windows. It is held together by 10s of
35:41
thousands of rivets. We need to transition from
35:44
single isolated holes to massive patterns.
35:49
What happens when holes start getting close to each other?
35:51
Do their individual stress shock waves collide?
35:54
They absolutely collide and they can either destroy each other or
35:57
save each other. Orientation is the deciding
35:59
factor. Let's start with the simplest
36:01
multiple hole scenario, 2 identical circular holes drilled
36:06
into a single flat plate. OK, two holes side by side,
36:09
maybe an inch apart. Right now, how are we applying
36:12
the load if you grip the top and bottom of the plate, pulling
36:16
perpendicular to the imaginary line that connects two holes?
36:19
Holes are sitting left and right and I'm pulling North and South.
36:21
Exactly. In this perpendicular scenario,
36:23
the narrow strip of solid metal residing directly between the
36:27
two holes is trapped. It has to carry a massive
36:30
unproportional share of the load because the stress lines from
36:34
both holes are being shoved into the same central bottleneck.
36:38
Double the bunching. The local stress shoots up
36:40
significantly compared to the baseline of a single isolated
36:44
hole. You have created a severe weak
36:47
link. That makes total intuitive
36:48
sense. You bottleneck the flow.
36:50
But now let's rotate the load. Take that exact same plate with
36:54
the holes left and right and pull it parallel to the line
36:57
connecting them. You pull the plate east and
36:59
West. OK, pulling in line with the
37:01
holes. An utterly counterintuitive
37:04
physical phenomenon occurs. The two holes actually shield
37:08
each other. Shield each other like race cars
37:10
drafting behind each other on a straightaway to cut wind
37:13
resistance. That is an exceptionally
37:15
accurate analogy. The 1st hole encounters the
37:18
primary stress flow lines, parting them, forcing them to
37:22
detour out and around its perimeter.
37:23
Right, pushing the flow wide. But before those parted stress
37:26
lines have the physical distance to smoothly close back in and
37:29
return to their uniform flat plate state, they immediately
37:33
crash into the 2nd hole. So the strip of metal sitting
37:36
directly between the two holes is literally sitting in the
37:39
aerodynamic wake of the 1st hole.
37:40
Yes, it becomes another low stress dead zone.
37:44
Because of this drafting effect, the maximum stress concentration
37:48
on the extreme outside edges of those holes actually drops below
37:52
the multiplier of three that we see for a single lonely hole.
37:55
It goes down. The closer the holes get to each
37:58
other in this parallel orientation, the deeper the wake
38:01
and the better they shield each other.
38:03
That completely upends how you think about removing material.
38:06
You're telling me that drilling a second hole right next to my
38:09
first hole can actually lower the peak stress on my part?
38:12
Welcome to the beautiful counter intuitive world of interactive
38:16
stress fields. Sometimes adding a void relieves
38:20
a bottleneck. Which naturally brings me to
38:22
industrial applications. If two holes interact, what
38:25
about 2000? I'm thinking of massive
38:26
infrastructure components, shell and tube heat exchangers and
38:29
refineries, or the colossal core support plates sitting inside
38:32
nuclear reactors. Massive grids of holes.
38:35
These thick slabs of steel are perforated with incredibly dense
38:39
grids of holes. How does an engineer even begin
38:41
to map the stress when the entire part is more hole than
38:44
metal? When dealing with densely
38:46
perforated plates, calculating individual interaction effects
38:49
is impossible. We zoom out and define a macro
38:53
level engineering metric called ligament efficiency.
38:57
Ligament efficiency breakdown The mechanics of that ratio the.
39:00
Ligament is simply the thin bridge of solid material
39:03
remaining between any 2 adjacent holes.
39:05
The metal between them. The efficiency is a simple
39:08
mathematical ratio, the minimum physical distance of that solid
39:12
ligament divided by the total pitch, the distance between the
39:16
exact center points of those same two holes.
39:18
So if the holes are massive and nearly touching, the solid
39:21
ligament is microscopic and your efficiency drops towards 0%.
39:25
If the holes are tiny and spaced 10 feet apart, the ligament is
39:28
huge and the efficiency approaches 100.
39:31
Percent exactly, and the massive manuals of stress concentration
39:35
factors for these reactor plates are all plotted linearly against
39:39
this ligament efficiency. But the absolute most critical
39:42
factor isn't just the spacing, it's the geometry of the grid.
39:45
The pattern. Whether the designer chooses to
39:48
arrange the holes in a square grid, a triangular grid, or a
39:51
diamond pattern entirely changes how the physical load navigates
39:55
the maze of ligaments. Is there a clear winner?
39:58
Does one pattern just flow better?
40:00
It is entirely dependent on the primary load vector, but
40:04
generally speaking, for states of equal biaxial tension, which
40:07
is exactly what you find in pressurized reactor place,
40:10
triangular hole patterns are vastly superior to square
40:14
patterns. Why?
40:15
Because a square grid creates straight continuous lines of
40:18
weakness. Partly, yes.
40:20
A square grid creates severe bottlenecks where the
40:22
perpendicular ligaments intersect.
40:24
But a triangular staggered pattern optimizes the stress
40:27
flow because it provides the load with continuous diagonal,
40:30
uninterrupted pathways to travel down.
40:33
Diagonal paths. The stress can weave through the
40:35
staggered ligaments much more gently, avoiding the harsh right
40:39
angle bottlenecks of a square grid.
40:41
Wait, you mentioned nuclear reactor plates.
40:42
Those aren't just holding a static weight.
40:44
They are heating up, cooling down, pressurizing, and
40:47
vibrating continuously for 40 years.
40:49
The load is cyclic. What a part full of interacting
40:52
holes is subjected to constant cyclic loading?
40:55
How do these stress multipliers translate to fatigue life?
40:59
This raises an important question, and you have just
41:01
touched on perhaps the single most critical warning I can
41:04
issue in this entire deep dive. Cyclic loading introduces the
41:08
terrifying reality of fatigue failure.
41:12
Fatigue is a killer. When an engineer is estimating
41:15
the fatigue strength for a perforated part, they must
41:18
calculate a reduced fatigue limit by applying a material
41:20
property called the notch sensitivity factor, denoted by
41:23
the letter Q. OK, Q tells us how sensitive the
41:26
specific metal is to having a hole in it under vibration.
41:28
Correct, but here is the fatal trap.
41:31
When applying Q to your equations, you must apply it
41:33
specifically to the net stress concentration factor KTN.
41:38
You absolutely, unequivocally cannot apply it to the gross
41:41
factor KTG. Let me push back on that to
41:43
understand the severity. What is the actual consequence
41:46
if a tired engineer at 2:00 AM accidentally plugs KTG into
41:51
their fatigue software instead of KTN?
41:53
Is it just a slightly less conservative safety margin?
41:56
No, it is much worse than that. Let's run a thought experiment.
41:59
Imagine you are designing a steel I-beam for a bridge and
42:03
you drill a massive row of holes down the center of the the web
42:07
to lighten it. OK.
42:08
So it's cheese beam? The baseline steel has a fatigue
42:11
limit of, say, 30,000 PSI before it cracks.
42:14
If you mistakenly use the gross factor KTG to calculate your new
42:19
fatigue limit, your mathematics will produce an absurd,
42:22
physically impossible result. How absurd.
42:24
The math will erroneously tell you that your Swiss cheese beam
42:27
actually has a higher fatigue limit than a solid uncut piece
42:30
of steel. It will tell you the holes made
42:32
the metal immune to vibration. Wait, the math will output a
42:35
number higher than 30,000 PSI. It will literally lie to you and
42:38
say removing metal made it physically stronger under cyclic
42:42
load. Yes, and the reason for the lie
42:44
is double counting the gross factor KTG already
42:49
mathematically bakes in the effect of the lost
42:52
cross-sectional area. It averages the stress over the
42:55
whole imaginary uncut plate. Right, the fantasy math.
42:58
By applying the notch sensitivity cue to KTG, you are
43:02
effectively double counting the missing material.
43:05
This mathematical error artificially inflates the
43:08
perceived strength of the remaining ligaments.
43:10
That's a huge blind spot. Your software will output a net
43:14
fatigue limit of say 38,000 PSI. You will certify the bridge for
43:18
Hwy. traffic believing you have a massive safety margin, but the
43:22
true net fatigue limit at the localized bottlenecks might only
43:25
be 15,000 PSI. Half as strong as you thought.
43:28
The bridge will endure a fraction of its design life,
43:31
cyclic fatigue cracks will rapidly initiate at the net
43:33
ligaments and the structure will catastrophically fail.
43:37
That is a chilling mathematical illusion.
43:39
You think you have a 20% safety buffer but you actually have a
43:41
50% deficit. Design junkies tattoo that on
43:44
your forearms. Always use KTN for fatigue
43:48
calculations in pattern elements.
43:50
Now I want to pivot because everything we have discussed so
43:53
far, from the infinite plates to the de Havilland comet windows
43:56
to the reactor grids, assumes we're dealing with relatively
44:00
thin sheets of metal 2 dimensional problems.
44:02
Right, plain stress. But my background is in heavy
44:06
machinery. I deal with massive thick cubic
44:09
blocks of hydraulic steel and my sketical brain is asking, does
44:14
the stress multiplier behave differently when it has to
44:18
travel down the deep dark barrel of a three inch long hole?
44:21
Your skeptical brain is touching on the frontier of
44:24
three-dimensional elasticity. For decades, engineers assumed
44:28
that the 2D plane stress solutions we just discussed
44:31
applied uniformly all the way down the bore of a thick.
44:33
Part. That would be the easy
44:34
assumption. But researchers Eli Sternberg
44:36
and MA Sadowski conducted rigorous 3D elastic analysis
44:40
that shattered those two-dimensional assumptions.
44:43
They delivered a revelation about thick plates that
44:45
completely upended forensic failure analysis.
44:47
Lay the 3D reality on us. In a thick element subjected to
44:51
tension, the maximum stress is not uniform down the barrel.
44:54
With the transverse hole, it varies wildly.
44:56
OK, so it changes. My intuition says the stress
44:59
must be highest right at the top and bottom exterior surfaces of
45:03
the block where the metal meets the air.
45:05
That's where the material is least supported.
45:07
That is the universal intuitive guess, and it is completely
45:11
wrong. Really.
45:12
Sternberg, Bergen, Sadowski proved that the stress is
45:15
actually slightly lower at the exterior surfaces.
45:18
It reaches its absolute peak multiplier slightly inside the
45:21
interior of the hole. Below the surface.
45:24
As you move from the visible surface down into the dark bore,
45:27
the stress climbs to a maximum localized peak before finally
45:31
settling into a uniform distribution deeper inside.
45:34
That defies common sense. Why would the metal trap deep
45:38
inside the whole experience worse stress than the exposed
45:42
metal on the surface? It comes down to the Poisson
45:44
effect and material constraint. When you stretch a block of
45:47
metal longitudinally, it naturally wants to contract or
45:50
neck down laterally. Like stretching a rubber band.
45:53
The metal right at the exterior surface is free to contract.
45:57
It has nothing but empty air pushing back against it.
46:00
That physical deformation relieves some of the localized
46:02
stress. But the metal buried half an
46:05
inch down inside the bore. Exactly.
46:07
The interior metal is completely surrounded and trapped by the
46:11
bulk mass of the surrounding steel.
46:13
It wants to contract laterally due to the Poisson effect, but
46:16
the surrounding metal fiercely resists it.
46:20
It's locked in place. This constraint creates a
46:22
complex state of triaxial tension.
46:25
The material is being pulled in three directions at once,
46:27
preventing it from yielding or deforming to relieve the primary
46:31
load. This triaxial constraint drives
46:33
the von Mises equivalent stress to its absolute maximum slightly
46:37
below the surface. Think about how utterly
46:39
terrifying that is for a maintenance technician.
46:42
If a hydraulic block is nearing the end of its fatigue life, a
46:46
tech might shine a flashlight on the exterior surface of the
46:48
hole, see absolutely 0 cracking and sign off the part is safe.
46:53
It's a false sense of security. Meanwhile, due to that triaxial
46:56
constraint, a microscopic fatigue crack has already
46:59
initiated invisibly down inside the barrel of the bore,
47:03
propagating outward where no one can see it and though the block
47:05
violently splits in half. It happens constantly in heavy
47:09
industry. It is a prime example of why 3
47:11
dimensional elasticity cannot be ignored.
47:14
The standard 2 DKT factors from the textbooks are usually
47:17
mathematically conservative enough to ensure your basic
47:20
design survives. But for forensics?
47:22
For forensic failure analysis, when you are trying to find
47:26
where the murder weapon is hidden, you must understand that
47:29
the crack almost certainly initiated deep inside the hole,
47:32
not on the convenient visible surface.
47:35
OK, what about holes that intentionally alter the surface?
47:38
I'm thinking of countersunk holes.
47:40
Earlier we talked about aerospace skins needing to be
47:42
aerodynamically perfectly smooth.
47:44
Yes, countersunk rivets. So they machine a conical divot
47:47
into the top of the hole so the rivet head sits flush.
47:50
Where is the stress peaking in a complex 3D shape like that?
47:53
Extensive finite element simulations reveal a very
47:57
distinct failure point for countersunk holes under tension.
48:00
The maximum stress does not occur on the smooth slope of the
48:03
cone, nor does it occur deep in the straight bore.
48:06
It occurs exactly at the sharp root.
48:08
The root being that specific transition line where the angled
48:12
cone of the countersink terminates and meets the
48:15
straight vertical Shank of the main bore.
48:17
Correct. It is a harsh sudden change in
48:20
internal geometry, and the stress flow lines bunch up
48:23
violently exactly at that transition line, specifically at
48:27
the flanks 90° away from the direction of the pull.
48:30
Right, the east and West flanks. And the data reveals a highly
48:33
counterintuitive detail. Changing the angle of that
48:37
countersink cone does practically nothing to help you.
48:40
Varying it from a steep 90° cone to a shallow 100° cone barely
48:45
nudges the stress concentration multiplier.
48:48
So if I have a failing design, I can't just tell the machinist to
48:51
tweak the angle to make it gentler?
48:53
What actually controls the failure risk?
48:55
Depth. The failure risk is strictly
48:57
driven by the depth of the countersank relative to the
49:00
total thickness of the plate. How much metal you leave behind.
49:04
If you carve away too much of the plate structural thickness
49:06
to sink the rivet head, leaving only a thin sliver of straight
49:10
Shank, the stress at that sharp root multiplies exponentially.
49:14
You are effectively severing the load path.
49:17
Which brings us to perhaps the most horrifying 3D stress
49:20
scenario imaginable, the absolute nightmare of fluid
49:24
dynamics engineers. Cross bores.
49:27
Let's visualize what a crossbore actually is for the audience.
49:30
Picture a massive solid block of forged steel forming the high
49:34
pressure head of an industrial compressor.
49:37
You drill a main bore straight down the center to carry fluid
49:40
at immense pressure. But the fluid needs to go
49:43
somewhere else, so you drill a secondary intersecting tunnel
49:46
straight through the thick side wall of the block, crashing
49:49
directly into the main bore at a perfect dead on 90° angle.
49:53
That intersection point is a crossbore.
49:55
You have two internal cylinders intersecting at a hard right
49:58
angle, buried inside a block of steel packed with pressurized
50:02
fluid trying to burst out. That sounds like a mathematical
50:04
disaster area. It is arguably the most severe
50:07
year localized 3D stress concentration you will routinely
50:11
encounter in mechanical design. The fluid pressure is
50:14
aggressively trying to pry that sharp 90° internal corner wide
50:18
open. I bet is the multiplayer.
50:20
A researcher named WF Riley conducted some incredibly
50:23
precise 3 dimensional photoelastic tests on these
50:27
exact crossbore intersections. He froze the stress states into
50:31
plastic models and sliced them open for his specific test
50:34
geometries. He found that the maximum
50:36
nominal stress factor hovering right at that sharp internal
50:39
crotch was around 3.6. But wait, earlier you said the
50:43
baseline for a circle was 3, so 3.6 is bad, but it doesn't sound
50:46
like a total disaster. But that's just the nominal
50:49
factor for his specific finite test block.
50:51
Right, exactly. His block had outer boundaries
50:54
that interfered with the pure theoretical multiplier if you
50:57
take Riley's empirical data and extrapolate it out to estimate
51:00
the true theoretical infinite body factor.
51:03
The unadulterated geometry. The pure geometric multiplier
51:06
without the outer walls of the block softening the blow.
51:08
The estimated theoretical KT sits at an astonishing 4.7.
51:13
Nearly five times the nominal stress localized at a
51:16
microscopic sharp internal corner buried in the dark where
51:20
you can never inspect it. How does an engineer even fix
51:23
that? You can't just leave a
51:24
multiplier of five inside a 10,000 PSI compressor block.
51:27
You can't. If you leave it sharp, it will
51:30
rapidly fatigue and violently burst.
51:33
The ultimate engineering fix to survive a crossbore involves a
51:36
rigorous 2 step mitigation strategy.
51:39
First, whenever physically and hydraulically possible, you must
51:43
make the two intersect and holes the exact same diameter.
51:46
Why does equal size matter? The finite element mathematics
51:49
and physical strain gauge measurements decisively prove
51:53
that unequal diameter, say a small pilot hole crashing into a
51:56
massive main bore, create much sharper, highly asymmetric
51:59
stress concentrations at the junction.
52:01
Equal diameters balance the disruption.
52:03
OK, that's step one. The second step is much harder.
52:06
You must physically reach deep inside that blind intersection
52:09
and completely round off the sharp internal corner where the
52:11
two cylinders meet. You have to machine a smooth
52:14
radius into an interior corner you can't even see.
52:17
Yes, you must use specialized radiusing tools to carve away
52:21
that sharp 90° Cliff. By creating a smooth sweeping
52:26
transition radius, you physically force the stress
52:29
lines to take a gentle sweeping path around the corner instead
52:33
of crashing into a dead end. Those are work if you execute
52:36
that internal radius perfectly. The mathematics show you can
52:40
theoretically drop the localized KT from near 5 all the way down
52:44
close to 1, but from a manufacturing standpoint it is
52:47
notoriously difficult and expensive to machine a perfect
52:50
Polish radius deep inside a blind intersection.
52:54
The classic battle between what the physics demand and what the
52:57
machine shop can actually physically build.
52:59
Which brings us perfectly to our final problem, the ultimate
53:02
practical compromise. Sometimes you just have to drill
53:04
the hole. I am an engineer.
53:06
I am forced by the assembly layout to put a hole right in
53:09
the middle of a highly stressed panel.
53:11
I don't have the space or the money to change the shape to a
53:14
beautiful optimized Inglis ellipse.
53:16
I am stuck with a standard circular drilled hole.
53:20
How do I physically reinforce that circle so my part doesn't
53:22
snap in half? If you are trapped with a circle
53:25
in a high stress field, you must utilize the art of the
53:28
reinforcement bead. You intentionally manufacture
53:32
the part with a thickened integral ring of solid material
53:35
completely surrounding the whole, much like the thickened
53:38
grommet around a shoelace hole. Just beefing up the perimeter by
53:42
adding a doughnut of extra metal.
53:43
Conceptually, yes, but mechanically you completely
53:47
alter the stress landscape when you add that thick and
53:50
reinforcement ring. You solve one problem but create
53:52
another. You no longer have just one
53:54
single failure point to worry about.
53:56
You now have to rigorously analyze 2 distinct competing
54:00
locations. OK, what is the first location?
54:03
The first is the exact inner edge of the hole itself.
54:07
Just like the baseline flat plate, you add a material so the
54:10
net stress is lower, but the multiplier effect of the circle
54:13
is still sitting right there at the boundary.
54:15
Makes sense. And the second location.
54:18
The second location is the outer perimeter junction, the exact
54:21
line where your thin flat panel abruptly steps up and meets the
54:24
thick wall of the reinforcement ring.
54:26
Ah. Because that sudden scare step
54:29
change in thickness is itself a geometric obstacle.
54:32
The stress flow lines flowing through the thin panel suddenly
54:35
smashed into a wall of thick metal that's going to cause its
54:38
own bunching effect. Exactly.
54:40
You are essentially trading one massive concentrated stress
54:44
point at the hole for two moderately smaller stress
54:48
points, one at the hole, one at the step.
54:51
And because the stress fields at that thickness junction are
54:53
highly complex and acting in three dimensions, we can't just
54:56
use simple 2D multiplier. We have to combine them.
54:59
We have to evaluate them using the von Mises equivalent stress
55:02
formula, which takes all the competing principal stresses
55:06
pulling in different directions and combines them into one
55:08
reliable digestible failure metric.
55:11
OK, so if adding a thicker ring makes the hole stronger, my
55:15
engineer brain immediately takes that to the logical extreme.
55:18
Does making the reinforcement ring wider and wider, stretching
55:21
it out like a massive frisbee and making it thicker and
55:23
thicker, just keep making the part infinitely stronger?
55:26
Can I just brute force the physics with enough mass?
55:29
I smile every time I hear that specific assumption because the
55:32
photoelastic data reveals A harsh, undeniable reality.
55:37
An emphatic no, you cannot brute force the physics.
55:41
There is a strict mathematical law of diminishing returns when
55:45
adding mass. Explain the physics of why more
55:47
metal stops helping. Let's say you cast a
55:50
reinforcement ring that is incredibly wide.
55:53
The diameter of the bead is five times the diameter of hole.
55:56
When researchers load that massive ring and view it under
55:59
polarized light, the photoelastic data clearly shows
56:02
that the outermost fringes of that incredibly wide ring
56:05
actually become photoelastically dead.
56:07
Just like the corners of the rounded square, they carry
56:10
absolutely zero stress. Exactly the same mechanism.
56:13
The stress flow lines moving at the speed of sound through the
56:15
steel simply do not have the physical time or geometric
56:19
inclination to migrate all the way out to the extreme outer
56:22
edges of your massive ring before the hole is already
56:24
behind. Them they just take the
56:25
shortcut. They take the shorter, tighter
56:27
path near the center. So you have bolted on all this
56:31
massive heavy material and the physics are actively ignoring
56:34
it. It is doing absolutely nothing
56:37
to carry the load. It is just parasitic deadweight.
56:40
In an aerospace or automotive design, parasitic deadweight is
56:43
the ultimate enemy. It burns fuel and ruins payload.
56:47
So if a massive frisbee is useless, what is the optimum
56:50
ratio? How do I tune this geometry to
56:53
get the absolute maximum strength for the absolute
56:56
minimum added weight? Through decades of testing, the
56:59
mathematical sweet spot has been clearly identified.
57:02
To achieve optimum efficiency, the ideal total thickness of
57:05
your reinforcement ring should be roughly 2 to three times the
57:08
baseline thickness of your panel.
57:09
Let me visualize that if my primary aluminum skin panel is 1
57:13
inch thick, my reinforcement ring should protrude so the
57:16
total thickness of the hole is between two and three inches.
57:19
Yes, that specific two to three ratio provides enough mass to
57:23
significantly blunt the stress bunching without entering the
57:26
zone of dead weight. The performance curves
57:29
demonstrate that the absolute largest, most dramatic gains in
57:32
reducing your KT happen with relatively tiny initial
57:36
additions of volume. Small changes, big results.
57:39
If you add just a little bit of smart mass, you can drop a
57:42
terrifying KT of 3.0 down to a highly manageable 1.5 very
57:47
efficiently. But.
57:48
What if I am an absolute perfectionist?
57:51
What if I want a KT of perfectly 1.0?
57:54
I want to add enough metal so the hole becomes completely
57:57
invisible to the stress field, as if it was never drilled at
57:59
all. The curves of the graph
58:01
asymptote violently. Trying to force your KT
58:03
perfectly back down to 1.0 is a fool's errand.
58:07
It requires a massive, heavy and highly inefficient volume of
58:10
material. You'd have to make the ring
58:11
enormous. To fight for that last 10% of
58:14
stress reduction, you have to add an exponential amount of
58:17
mass. For aerospace engineers, that is
58:19
almost always a catastrophic tradeoff.
58:22
A smart designer readily accepts a KT of 1.2 or 1.3, mitigating
58:27
the peak danger in order to save hundreds of pounds of payload
58:30
over the entire aircraft. You compromise with the physics,
58:34
you don't try to defeat them entirely.
58:36
Wow. OK, we need to land the plane of
58:38
this master class. We have been on an incredible
58:40
journey today. We started with the violent
58:43
reality of a microscopic pinhole tripling the stress in a steel
58:47
continent. We explored how pulling in all
58:50
directions at once can actually save a pressure vessel, reducing
58:53
the factor to two. It's been a ride.
58:55
We dug into the tragic history of square airplane windows and
58:58
the mathematical elegance of the two to one vertical ellipse that
59:01
solved it. We ventured into the terrifying
59:03
invisible 3D stresses lurking inside cross pores, and we
59:07
finally established the exact volumetric limits of how much
59:10
metal you should add to reinforce a simple circle.
59:13
It is a dense, mathematically heavy topic, but if you step
59:17
back, it all comes back to a single philosophy respecting how
59:20
force desires to flow through physical geometry.
59:24
We don't dictate the load path, we merely provide the roads.
59:28
But I want to leave you and the listeners with a final paradigm
59:31
shift to ponder as you look at the world.
59:33
I am always ready for a paradigm shift.
59:35
We spend virtually all of our time in mechanical engineering
59:39
trying to mitigate, reinforce, and apologize for the damage a
59:42
hole does to our pristine stress fields.
59:45
We view voids as the enemy. But I challenge you to look at
59:48
nature. Look at the cross section of a
59:50
human femur bone, the grain structure of Oakwood, the
59:53
microscopic scaffolding of bamboo, right?
59:55
They are not solid bars of material.
59:57
They are fundamentally entirely constructed out of microscopic
1:00:00
holes. There are massive cellular
1:00:02
structures entirely made of voids.
1:00:04
Yes, and nature does not view those empty spaces as flaws.
1:00:08
Nature uses them deliberately. It doesn't just use them to save
1:00:11
weight. It intentionally places specific
1:00:14
shapes of holes to arrest propagating microcracks and to
1:00:18
softly, actively route catastrophic stress waves around
1:00:22
the mital delicate areas of the Organism.
1:00:25
The empty space is doing the heavy lifting.
1:00:27
The next great frontier of mechanical engineering isn't
1:00:30
just learning how to bolt on enough metal to survive the
1:00:33
holes we drill. The frontier is learning how to
1:00:35
place voids intentionally creating smart geometries where
1:00:39
the nothingness actively mathematically guides the force
1:00:43
exactly where we want it to go. That is a profoundly beautiful
1:00:46
way to look at it. The void isn't the absence of
1:00:49
design, it is the ultimate tool of design, designing the empty
1:00:52
space to be just as functional, just as vital as the solid
1:00:55
material surrounding it. Remember, design junkies master
1:00:58
the critical few or the trivial many will master you.
1:01:01
If you're leaving this eisode with something you didn't arrive
1:01:03
with, drop a comment. Takes you a few seconds, but it
1:01:05
tells me this workbench is worth keeping open.