Latest / Mechanical Engineering Made Simple / (#69) Reliable Gear System Design Principles

Transcript

(#69) Reliable Gear System Design Principles · Mechanical Engineering Made Simple

0:00
Welcome back design junkies. If you enjoy the podcast, let us
0:02
know the like or follow and get ready for Mechanical Engineering
0:06
Made Simple. Simple perhaps, but certainly
0:09
not shallow. We are taking a deep dive into
0:12
mechanical engineering principles today, focusing
0:14
specifically on reliable gear system design.
0:18
Absolutely not shallow. You, our listener, have handed
0:21
us a stack of sources that I mean they represent the deepest
0:25
cuts of gear, engineering theory and practice.
0:28
They really do. So our mission today is to
0:30
provide you with the ultimate shortcut.
0:32
We are tackling the most pressing, complex questions that
0:35
keep experienced engineers up at night.
0:37
Things like how to size gears correctly, how to quiet them
0:40
down. And critically, how to prevent
0:42
catastrophic failure in the field?
0:44
The goal here is to move well beyond, you know, the simple
0:47
concept of two meshing teeth and basic geometry.
0:50
As we dive in, we'll quickly see that designing A gearbox isn't a
0:53
singular calculation. No, it's a high stakes balancing
0:56
act. It really is.
0:58
We are simultaneously wrestling with, I don't know, two dozen
1:01
trade-offs balancing high torque requirements against high speed
1:05
controlling destructive heat buildup.
1:07
Suppressing that intrusive acoustic noise and making sure
1:10
the system survives under extreme dynamic multi axis
1:14
loads. It's a continuous battle against
1:16
real world messy physics. The kind of physics that
1:19
spreadsheets often ignore. Exactly.
1:22
So how are we structuring this deep dive to give you the most
1:25
logical path through all this complexity?
1:28
We will move step by step. We have to start logically at
1:31
the foundation fundamental sizing load and stress
1:34
calculation, which is, you know, the necessary bedrock.
1:37
OK. Then we'll pivot to the
1:38
essential safety net, material science, specialized heat
1:42
treatment and protective safety factors.
1:44
Got it. From there, we address the
1:46
system's dynamic behavior, noise, vibration, resonance
1:49
control. Then we'll zoom out a bit to
1:50
look at external influences like alignment and housing
1:53
flexibility before we conclude with the critical issues of
1:56
thermal efficiency, lubrication, and finally the definitive
1:59
process of failure analysis. Perfect.
2:01
OK, let's unpack this starting right at the fundamentals you
2:04
mentioned. How do I size gear for torque,
2:08
speed and duty cycle? And I guess most importantly,
2:12
how do I calculate tooth bending stress correctly?
2:16
Well, when you're sizing any gear, you're fundamentally
2:19
designing against two primary failure modes, and these are
2:23
codified in engineering standards, like for instance the
2:26
well known ANSI AGMA 2001 D 04. And those two modes are.
2:30
First, you have pitting resistance.
2:32
That's the surface contact fatigue that causes, you know,
2:35
small pieces to flake off the toothpaste.
2:37
Pitting. And second, you have to ensure
2:39
bending strength which is measured by the stress
2:42
concentration right of the tooth would fill it.
2:44
And the foundational math for that bending strength, the
2:47
classic starting point that every engineer learns.
2:50
That's the Lewis equation. It's the historical bedrock,
2:52
yes. The Lewis equation gives you a
2:54
fundamental measure of the tangent gentle tooth load, which
2:57
we symbolize with a W. It's really the capacity of a
3:00
spur gear tooth to withstand bending.
3:02
OK, so how is it written? For metallic spur gears, the
3:04
formula is written as W = S * S * y all divided by P and then
3:11
that whole thing is multiplied by the ratio of 600 over the
3:14
quantity 600 + V. Wow, that looks dense.
3:18
That looks like something ripped straight out of a textbook from,
3:20
I don't know, the 19th century. Can we break down the practical
3:23
meaning of those variables? Certainly.
3:25
So W is your static tooth load in pounds force right along the
3:28
pitch line, S is the safe material stress allowable, which
3:32
is typically the static stress in pounds per square inch.
3:34
OK, F is the face width in inches, P is the diametral pitch
3:38
which really just defines the size of the tooth, and Y is a
3:41
tooth form factor which accounts for the actual shape and
3:44
geometry of the tooth at its root.
3:46
So the equation basically takes the material strength,
3:48
multiplies it by the contact area, the face width and tooth
3:51
size, and then adjusts everything for the specific
3:55
shape of the tooth. Precisely.
3:56
But here is the critical limitation that adjustment term
3:59
that ratio of 600 to 600 + V It's an empirical attempt to
4:03
account for dynamic forces. Empirical meaning based on
4:06
observation, not pure theory. Exactly, and because of that,
4:10
it's only considered valid for pitch line velocities V up to
4:14
about 1500 feet per minute. 1500 feet per minute.
4:17
So what happens if you go faster?
4:19
Beyond that velocity, this simplified model just fails.
4:23
It fails catastrophically because the dynamic effects of
4:27
tooth to tooth impact, they become overwhelming.
4:30
And in modern high speed drives, like in an EV, you're way past
4:33
that. Way past it, which is why
4:35
engineers are forced to use much more complex iterative AGMA
4:39
methods that factor in quality and deflection rather than
4:43
relying on this historical approximation.
4:45
Wait, if the Lewis equation breaks down at 1500 feet per
4:48
minute, doesn't that make it sort of obsolete for a lot of
4:51
modern high speed applications? The limit is critical because
4:54
above that speed, the gear mesh no longer behaves like a steady
4:57
state load transmission. It becomes a series of high
5:00
frequency impacts, the inertia of the teeth, the slight
5:03
manufacturing errors. They create these acceleration
5:06
and deceleration impulses that are much larger than the
5:09
transmitted static. Load so it's still useful for
5:11
low speed high torque stuff. Absolutely.
5:13
For low speed industrial applications, it's still
5:15
foundational, but for everything else we move immediately into
5:19
dynamic factors and crucially, service factors.
5:22
And that brings us to the next big question.
5:25
How do we integrate the actual operating conditions, the
5:28
environment, the fluctuating load, the duty cycle into that
5:32
sizing decision? This is where service factors
5:34
come in and really affect gear life.
5:36
This is exactly where service factors step in.
5:39
They drastically scaled the required gear size and capacity.
5:43
They are the engineer's way of acknowledging uncertainty.
5:45
And you know all those external conditions.
5:47
So you're not just designing for the average load?
5:50
Never. You're designing for the worst
5:51
case scenario over the life of the machine.
5:54
We manage these real world variations by applying various
5:57
factors in the drive system selection process.
6:00
Can you give us a concrete practical example?
6:02
How does the duty cycle scale the required capacity?
6:06
Sure. Think about a conveyor bolt
6:07
drive in a factory. We classify the load based on
6:10
its shock and frequency. For instance, the low
6:13
classification might be medium impulsive, often labeled F sub D
6:16
Then you apply the starts per hour factor F sub S So if you
6:21
have a drive system that starts and stops say 15 times per hour,
6:25
you have to apply a starts per hour factor of 1.2 to your
6:28
calculation. A 20% increase.
6:30
Right, but if that same system started 150 times per hour, that
6:34
factor would jump significantly higher, maybe to 1.5 or 1.6
6:38
depending on the standard you're using.
6:39
So that factor, that factor directly scales the necessary
6:43
design capacity. You might calculate you need a
6:46
10 horsepower gearbox, but after applying that 1.2 factor you
6:50
have to select a 12 horsepower rated gearbox just to handle
6:53
those frequent transient startup loads.
6:55
Exactly. You are proactively designing
6:57
for the transient peak loads, which often include inertial
7:00
forces and motor startup torque, not just the steady state load.
7:04
But even during continuous smooth operation, the tooth
7:06
contact isn't truly static, right?
7:08
We need to account for what happens when those, in theory,
7:11
highly accurate surfaces slam into each other thousands of
7:14
times a minute. Which brings us right back to
7:16
dynamic loads. Static strength calculations are
7:19
clearly not enough. Precisely running gear teeth
7:22
experience dynamic loads that must be accounted for, and these
7:25
loads are heavily influenced by maintaining close control on
7:29
gear tooth accuracy during manufacturing.
7:31
So if the teeth are sloppy. If the teeth are inaccurate or
7:34
poorly finished, the peak dynamic loads will be immense,
7:38
potentially exceeding the static strength capacity of the tooth
7:41
instantly. How do we quantify that dynamic
7:44
impact in the design phase then? One approximation for the
7:47
dynamic load F sub D is given by F sub D equals the ratio of 1200
7:53
+ v sub G to 1200, all multiplied by the actual
7:57
tangential load. F&V sub G is the pitch line
8:00
velocity of the gear. Correct.
8:02
And the critical design rule is simple.
8:04
This calculated dynamic load must never exceed the endurance
8:07
load capacity of the tooth material.
8:08
So how is the gear's inherent quality?
8:11
It's precision grade factored into controlling that dynamic
8:13
load. This feels like we're
8:15
manufacturing quality directly impacts the physics.
8:18
That's handled by the dynamic load factor K sub V.
8:20
This factor is absolutely crucial in design.
8:22
It provides the link between geometry, speed and precision.
8:25
OK, So what is it? You can think of a gear mesh
8:27
like a car driving over a road. If the gear is low precision,
8:31
it's like driving over railroad tracks.
8:33
Bumpy, high impact, lots of momentary peak forces.
8:37
But if the gear is high precision.
8:39
It's like driving on a perfectly smooth, newly paved highway.
8:41
So case of V numerically quantifies that bumpy Rd.
8:45
penalty. Exactly.
8:47
The higher the precision grade, like an AGMA 14 or 15, the
8:51
closer to one point O this factor gets, minimizing the
8:54
penalty. High speed paired with low
8:56
precision means a very large case of V which significantly
9:01
dedrates your permissible. Static load, so it's often
9:03
cheaper to just specify a higher quality gear.
9:06
Much cheaper and safer than simply trying to use a stronger
9:09
but lower quality material. So if the dynamic loads are too
9:12
high, the engineer has two paths, increase the static
9:15
strength through material or size or decrease the dynamic
9:19
penalty by increasing the quality grade through
9:21
manufacturing precision. That's the choice.
9:24
Let's follow that first path for a minute and talk about
9:26
materials, which determines that safe allowable stress S.
9:29
We've established that the materials safe allowable stress
9:33
S is a primary variable. But before we get into specific
9:37
materials, let's address a foundational safety question
9:40
that often puzzles designers. What safety factor should I
9:43
actually use for gears? It feels like a rule of thumb,
9:46
but you mentioned it's far more nuanced.
9:48
It is the required safety factor which we often denote as S sub H
9:53
for Hertzian contact. Stress is highly dependent on
9:56
where the failure initiates. Where it initiates, yes.
9:59
And here's a piece of advanced insight that separates expert
10:01
designers from novices for gear pairs that have a large number
10:06
of teeth, meaning a larger pitch diameter.
10:08
For a given tooth size, the point of maximum shear stress
10:11
occurs deep beneath the surface of the gear flank.
10:14
Beneath the surface, why does the geometry push the stress
10:16
deeper into the material? It relates to the shade and size
10:19
of the contact zone, the Hersen contact patch.
10:22
As that contact area grows the peak stress zone, which is
10:26
critical for failure, it moves away from the surface.
10:28
So if that maximum shear stress point occurs beyond the effect
10:31
of case depth, that thin hardened wear resistant layer
10:35
from heat treatment. Then the gear is at high risk of
10:38
a deep subsurface failure. It's often called case crushing.
10:41
That completely redefines how you specify the alloy.
10:44
The take away is if I'm designing a huge wind turbine
10:47
gear with 200 teeth, I should look immediately at the
10:50
materials core strength, not just a surface hardness.
10:53
Precisely because the stress is peeking deep inside in these
10:58
specific high tooth count, high load cases, the engineer must
11:02
deliberately use a larger safety factor S sub H to ensure the
11:06
relatively softer core material can handle that load, preventing
11:10
the hardened case from collapsing prematurely.
11:12
Speaking of material treatment, let's discuss that fundamental
11:15
trade off. What gear material should I use
11:17
for high load versus high speed? We move from specialty steels to
11:21
plastics depending on the constraints.
11:22
For high load, high stress applications, you have to use
11:25
specialized steel alloys. Standard carbon and alloy steel
11:28
shafting typically falls in the Rockwell C hardness range of,
11:31
say, 40 to 58. But for the really critical
11:34
stuff. For the most critical
11:36
applications like auto or aircraft engine shafts that face
11:39
extreme cyclic fatigue, you require high strength speciality
11:43
alloys, things like E4130U4140 and E4340.
11:49
These are chosen specifically for their ability to maintain
11:51
strength and integrity under extreme stress and heat cycling.
11:55
And how does that trade off shift when we talk about high
11:58
speed, where friction and thermal management become the
12:00
dominant problem? Let's consider the worm gear,
12:03
which is fundamentally a low efficiency device because of its
12:06
inherent high sliding contact, the allowable stress factor S
12:09
sub climb. It varies dramatically based on
12:12
the material pairing. So what's the best pairing?
12:15
If you pair phosphor bronze, specifically a centrifugal
12:18
casting which gives you superior density, with an alloy steel
12:21
that has been carburized and quenched, you achieve the
12:23
highest speed limit, approximately 30 meters per
12:25
second. That pairing excels at handling
12:27
the friction and the heat. And if you choose a lower grade
12:30
material like brass. Contrast that with using brass,
12:33
which might be limited to only 8 meters per second in a similar
12:36
service. This huge difference exists
12:39
because high speeds and sliding contacts are primarily limited
12:43
by the high temperatures generated by friction.
12:46
So the material choice dictates the thermal ceiling.
12:49
And thus the maximum permissible speed before the lubricant film
12:53
breaks down or the material itself just softens.
12:57
That thermal limitation brings us to the other end of the
12:59
material spectrum, plastics, materials that trade strength
13:03
for low friction, low cost and noise reduction.
13:07
Right materials like MC nylon and acetyl copolymer.
13:10
They're used widely because they're low cost, low friction,
13:13
and very lightweight. They often serve in small
13:16
precision applications like office equipment or consumer
13:18
electronics. There's a catch.
13:20
There's a big catch. Their use introduces a critical
13:22
dimensional control challenge that Metal Gears just don't face
13:25
to the same extent. Which is.
13:27
These plastic materials have large coefficients of thermal
13:30
expansion and they tend to absorb moisture.
13:32
Sometimes they can swell by 1% of their size and high humidity.
13:35
1% is massive in gear terms. It is.
13:39
It requires the designer to allow for a significantly
13:41
greater backlash allowance than they would for Metal Gears.
13:45
If you fail to account for that swelling, the gear train will
13:48
lock up, generate immense force, or suffer catastrophic
13:51
interference. You're essentially designing for
13:53
the worst case dimension in the harshest environment.
13:56
So plastics trade strength for noise and cost, but introduce
14:00
these huge dimensional control problems.
14:03
Let's get back to those high strength steel gears and address
14:06
the heat treatment process. How hard is too hard for a gear
14:10
tooth? Well, the entire purpose of heat
14:13
treatment, particularly carburizing and quenching, is
14:16
strategic. It aims for balance.
14:18
A balance between what? You're trying to boost surface
14:20
strength, creating a very hardware resistant surface that
14:24
could resist pitting while at the same time preserving A
14:26
ductile core underneath to handle bending fatigue without
14:30
an immediate brittle fracture. It's a gradient of hardness.
14:33
How do engineers define and measure that strategic balance?
14:36
They look at the effect of case depth.
14:38
This depth is formally defined as the distance into the
14:41
material where the hardness is greater than the high 513, which
14:45
is approximately HRC 50 on the Rockwell C scale.
14:48
And that depth has to be controlled.
14:50
Precisely controlled for ground gears where material has been
14:53
removed after hardening. The effective case death is
14:56
specifically the residual layer death that remains after
15:00
grinding to the final dimensions.
15:02
If that residual death is too shallow, the case integrity is
15:05
lost and the gear fails very rapidly.
15:09
And if the gear is large or complex, like a helical ring
15:11
gear, controlling the heat treatment process must be an
15:14
incredible challenge, especially avoiding warping or that unwind
15:18
distortion. Distortion control is paramount,
15:21
especially for precision ground parts.
15:23
Techniques like induction hardening are often used
15:25
specifically to localize the heat and achieve strict
15:28
dimensional tolerances. But the most critical stage is
15:30
the quench. The most critical stage is the
15:33
quench portion design. Factors like the cooling rate,
15:37
the precise design of the fixture holding the gear, and
15:40
the duration of the cooling process are all
15:42
deterministically critical for controlling crown and unwind
15:45
distortions in the gear tooth. Can you give an example?
15:48
Sure. Manufacturers of extremely large
15:50
wind curbine components, for instance, often use vacuum
15:53
carburizing followed by high gas pressure quenching.
15:56
As a modern, clean and highly controlled alternative to
16:00
traditional atmosphere carburizing and press quenching,
16:03
it leads to far superior dimensional stability.
16:05
The precision required in that quench process alone is just
16:08
astounding. OK, let's shift our focus now
16:11
from static strength and material integrity to the
16:13
system's dynamic behavior. The next major challenge is
16:17
ensuring the machine runs quietly, right?
16:19
So if a client asks why is my gearbox so noisy at certain
16:22
speeds, what's the 1st place the design engineer looks?
16:25
It's not usually the motor, is it?
16:26
They look at dynamic excitation. That's the immediate source of
16:30
noise, and while excitations can come from external mechanical
16:33
assemblies, motors, pumps, hydraulic cylinders, the most
16:37
crucial and often most insidious source is the gear quality
16:40
itself. So back to manufacturing.
16:42
Back to manufacturing, Inaccurate and poorly finished
16:46
teeth, excessive backlash. These are consistently
16:49
identified as major contributors to peak dynamic loads and the
16:53
resultant acoustic noise. They essentially excite the
16:56
entire casing like a drum. So poor manufacturing quality
17:00
translates directly into high frequency impact loads which
17:03
generate noise. But beyond simple tooth
17:05
imperfections, what's the primary vibrational source that
17:08
an engineer needs to model in the shafting?
17:11
The principal source of vibratory torques in the entire
17:13
system is torsional vibration. This vibration is excited by the
17:17
cyclical firing or pulsing of the power source, be it a piston
17:20
engine, pumps, or even the ripple torque from an electric
17:23
motor. And it travels through the
17:24
shafts. It's transmitted through the
17:26
shafting, causing the gears to excite the casing and the
17:29
surrounding environment. This torsional vibration causes
17:33
the most objectionable, often harmonic, gear noise.
17:37
To combat that noise, engineers use micro geometry adjustments.
17:41
Can you explain in simple terms how design features like tip
17:44
relief and crowning reduce gear noise?
17:46
Sure. Tip relief and crowning are
17:48
known collectively as flank modifications or micro geometry
17:52
adjustments. They're implemented during the
17:54
design and machining process. And what do they do?
17:57
They are design changes implemented precisely to
17:59
compensate for the tooth deformation that is expected
18:02
under the working load. When a gear tooth deflects under
18:05
torque, it changes the ideal point of contact and can lead to
18:09
sudden noisy impacts when a new tooth pair engages.
18:12
So you are basically pre denting the tooth geometry so that when
18:15
the load hits it straightens out perfectly and achieves smooth
18:18
contact. Exactly.
18:20
You've got it. These modifications
18:21
strategically alter the tooth surface so that as the load is
18:25
applied and the tooth deflects, the contact zone shifts smoothly
18:29
to the optimal quietest point. It ensures continuous
18:33
distributed load sharing and minimizes that instantaneous
18:37
impact. That's genius in its simplicity,
18:39
designing for the deformed state rather than the static one.
18:43
Are there other advanced techniques for surface structure
18:45
modifications that target specific noise harmonics?
18:48
Yes there are. For high precision applications,
18:52
specific grinding processes can intentionally alter the gears
18:55
surface structure along the path of contact.
18:58
The entire goal is to influence and improve the excitation
19:01
behavior. To get rid of a specific wine.
19:04
Exactly to address an objectionable harmonic noise,
19:07
for instance, specific surface shifts can be introduced to the
19:09
generated member of a ground bevel gear set to subtly alter
19:13
the contact kinematics and remove those nasty high
19:16
frequency wine components. That addresses the quality of
19:19
the tooth itself, but if the entire shaft or rotor system
19:22
begins to oscillate at a natural frequency, we hit resonance, and
19:25
that's a high stakes system destroying scenario.
19:28
Catastrophic. So how do we keep the system
19:30
from shaking itself apart? How do I avoid resonance in gear
19:35
systems? Resonance is catastrophic
19:37
because even small forces applied repeatedly at that
19:40
natural frequency can build up huge destructive amplitudes.
19:45
We mitigate this first through careful modeling.
19:47
You have to know where the danger zones are.
19:49
You must calculate the critical speeds, which are the natural
19:52
frequencies of the system. This is often done using complex
19:55
finite element methods or FEM for systems like ship propulsion
19:59
shafts or large rotating machinery.
20:02
Once those critical speeds are known, what's the primary
20:05
corrective action? You can't just tell the operator
20:07
to avoid that speed range. Well, sometimes you do, but the
20:11
primary action is shifting the critical speeds out of the
20:13
operating range. If you increase the stiffness of
20:16
the shafting in a system by a ratio, say alpha, then all the
20:20
frequencies will increase by that same ratio alpha, assuming
20:23
the inertia is constant. And there's a trick to it.
20:25
Here's the key design. Leverage changes in stiffness
20:29
made near a nodal point. A point on the shaft with
20:32
minimal displacement will have the maximum effect on shifting
20:36
the natural frequency. And for inertia.
20:38
Conversely, changes in inertia made near a loop a point of high
20:42
displacement will have the maximum effect.
20:45
So strategically placing stiffness or inertia changes
20:48
gives the designer leverage over the system's frequencies.
20:52
What if you can't shift the speeds enough due to, say,
20:55
packaging constraints? When you turn to active damping,
20:58
dampers of various types, like tuned viscous dampers can be
21:01
added. This is particularly necessary
21:04
when torsional resonances are calculated to fall within a
21:06
critical frequency range, typically ±10% of the limiting
21:10
operational frequency. And specifically for the gear
21:12
mesh itself in spur gear systems.
21:14
The operating strategy is it's non negotiable.
21:18
They must be operated either above or below the resonance
21:21
point, avoiding that peak frequency altogether.
21:24
This often means running a system at an artificially higher
21:26
speed or strictly limiting operation below a critical
21:29
range. Which is usually managed by the
21:31
control software. Since that destructive vibration
21:34
often originates with the engine or motor, how do we use
21:37
couplings to achieve torsional isolation?
21:40
Couplings are critical for decoupling the driven machinery
21:43
from engine excited torsional vibration.
21:47
Take the example of the TWI flex coupling, A specialized fluid
21:50
filled or flexible element coupling.
21:53
It achieves isolation if the ratio of the coupling stiffness
21:56
to the transmitted pork is kept below about 7:00 to 1:00.
21:59
If that ratio is maintained, the major machine modes driven by
22:03
the gearboxes are not excited by the source vibration.
22:07
The coupling itself acts as a sacrificial spring element.
22:11
It absorbs those high frequency pulses before they reach the
22:13
gear train. Precisely.
22:15
That's a perfect segue, because couplings often compensate for
22:18
issues the system can't fix internally.
22:20
Let's look at the external environment, the housing, the
22:23
shafts, the bearings, and how they impact gear performance.
22:26
We'll do it. So we've designed the perfect
22:29
tooth and solved our vibration problem.
22:32
Now let's talk about mounting. Improper mounting is a
22:35
notoriously frequent cause of poor tooth contact and premature
22:39
failure. How sensitive are gears to
22:42
misalignment? What are the big errors
22:45
engineers need to worry about? Gears are extremely sensitive to
22:48
misalignment. It destroys the entire premise
22:51
of the design that the load is distributed evenly across the
22:55
full face with misalignment concentrates the load, leading
22:58
to itting failure near the edges.
23:00
What are the main culprits? There are three major external
23:03
errors. First, the shafts are skew,
23:05
meaning they are nonintersecting and misaligned, 2nd, a shaft
23:09
angle error in the gearbox itself, and 3rd a mounting
23:12
distance error which changes the distance between the two shaft
23:15
centers. Let's focus on that mounting
23:16
distance error. It seems like the easiest one to
23:19
mess up during assembly. How does that error practically
23:22
affect the contact pattern on the tooth?
23:24
When you have a positive mounting distance error, meaning
23:26
the center distance is too large, it physically moves the
23:29
contact point on the pinion toward the tooth root.
23:32
And on the other gear. Conversely, the contact on the
23:34
meeting gear moves toward the tooth tip.
23:37
This has an effect similar to having a slightly smaller
23:39
pressure angle on the pinion, and it concentrates the stress
23:42
unevenly. And you can fix that.
23:44
For bevel gears, yes, these errors are often diminished
23:48
through field adjustment by adding or removing a backing
23:50
Shim to achieve the correct tooth marking test result.
23:54
That highlights the need for precision beyond just the gear
23:57
surface. We talked about gears and
23:59
shafts, but the performance is fundamentally limited by the box
24:03
they sit in. How do bearing stiffness and
24:06
housing deflection affect gears? How does a designer account for
24:10
all this flexibility? This requires an integrated
24:13
design approach. The gear has to be treated as
24:15
part of a flexible system, not a rigid component.
24:18
Housing deflection and bearing stiffness are absolutely not
24:21
isolated factors. They dictate the position of the
24:23
shafts. They determine the positional
24:25
accuracy of the shafts under load, which in turn dictates the
24:28
exact load distribution across the gear teeth.
24:30
Since the maximum capability of the gears relies on a uniform
24:33
load distribution, this flexibility is a massive
24:36
concern. So no more simple beam
24:38
calculations for the housing. We need advanced analysis.
24:42
Absolutely. For high performance
24:44
applications, engineers rely heavily on advanced Computer
24:47
Aided Engineering, or CAE and Finite Element analysis.
24:51
FEA complex models are built that include the flexibility of
24:55
all the components, the gears, the shafts, the bearings, the
24:58
casing structure. And this simulates how the
25:00
system deforms under load. Exactly.
25:03
It simulates the system deflection under full load,
25:06
predicting the resulting gear misalignments and most
25:08
importantly, the resulting contact patches.
25:11
The contact patch is the ultimate measure of success,
25:14
isn't it? It is.
25:15
These simulating contact patches are then compared to physical
25:18
tooth marking test results, where you literally paint the
25:20
teeth and run them to verify the accuracy of the complex FEA
25:24
model. If the simulation doesn't match
25:26
the test, the design is flawed. This issue has to be severely
25:29
compounded in planetary gear systems where you have multiple
25:32
planets trying to share the load.
25:34
In an ideal planetary set, every planet transmits the exact same
25:37
torque, but reality shows that manufacturing variations, the
25:41
subtle effects of gravity, and, critically, carrier bearing
25:44
clearance all result in unequal load sharing.
25:48
And there's data on this. There is empirical testing in
25:51
large industrial gearboxes has provided telling data.
25:54
Gearboxes equipped with standard cylindrical roller bearing saw
25:57
load sharing factors as high as 1.16.
26:00
Meaning one planet was carrying 16% more load than average.
26:05
A 16% overload on one component dramatically shortens the
26:08
lifespan of the entire gearbox. It does.
26:12
However, gearboxes using tapered roller bearings, which offer
26:15
better control over radial and axial positioning under load,
26:18
managed to keep that factor lower, closer to 1.10.
26:21
Which demonstrates that the type of bearing a seemingly small
26:25
decision has a measurable and crucial impact on the load
26:28
distribution. It proves the necessity of
26:30
integrated modeling. Let's swing back to couplings.
26:32
They are the forgiveness mechanism in the system.
26:34
They have to accommodate the errors we've just discussed.
26:37
Flexible couplings are essential for accommodating the inevitable
26:40
combinations of misalignment, angular parallel, offset and end
26:44
float, which is axial motion. Manufacturers offer various
26:47
types, chain flex, dis flex, gear flex, each capable of
26:51
handling specific degrees of these misalignments.
26:54
We mentioned hooks joints earlier in relation to
26:57
vibration. Why are specialized couplings
26:59
preferred over standard universal joints when constant
27:02
velocity is required? This is crucial, particularly in
27:05
sensitive drivetrain systems. A conventional hooks joint
27:08
inherently induces cyclic pulsations or variation in
27:11
rotational velocity. When it's operating under high
27:14
misalignment, the output speed actually speeds up and slows
27:17
down twice per revolution. Even if the input is perfectly
27:20
constant. Exactly.
27:22
Specialized flexible drive shaft couplings are preferred because
27:26
they are designed to transmit high torque and tolerate high
27:28
misalignment without that cyclic speed variation, ensuring smooth
27:33
constant velocity transmission. So we designed the gear itself.
27:36
Analyze the entire higher flexible housing and bearing
27:38
system and select couplings that compensate for the residual
27:42
external errors. Our next major headache is what
27:46
happens to the energy that isn't transmitted as useful power
27:49
friction and heat the. Hidden costs.
27:52
Efficiency is key, especially in modern EV powertrains where
27:55
every Watt counts. What causes sliding losses
27:59
versus rolling losses? And where else does the
28:02
efficiency drain occur in a gearbox?
28:04
Gearbox efficiency is reduced by several sources of power loss.
28:08
The most significant are elastohydrodynamic friction
28:11
losses, EHL friction losses in the tooth contacts themselves.
28:16
This sounds complex. It involves both sliding and
28:18
rolling action. The efficiency penalty comes
28:20
from the energy required to shear the oil film trapped
28:23
between the meshing surfaces. But beyond the tooth contact,
28:26
you have other significant contributors, like what bearing
28:28
losses, seal losses and churning losses.
28:31
Churning losses. The energy wasted simply
28:33
splashing or churning the oil around.
28:35
Especially in systems where the gears are submerged, this
28:37
becomes a major factor at high rotational speeds.
28:40
Which is why high speed systems use an oil mist.
28:43
Right. To minimize that drag, EHL
28:45
friction is the major focus for efficiency optimization.
28:49
To highlight the spectrum, conventional spur gears can be
28:51
97 to 99% efficient per stage. Worm gears, however, they're
28:55
much. Lower.
28:56
Much lower. They can dip below 50%
28:58
efficiency due to that high sliding component and that's why
29:02
new designs like parallel axis self locking gears are being
29:06
developed. They approach the efficiency of
29:08
conventional gears and critically, generate
29:11
significantly less heat. Now let's tackle the symptom
29:15
every engineer dreads. Why is my gearbox overheating?
29:18
And how do I estimate heat generation in gears?
29:21
Overheating is simply a failure of thermal balance.
29:23
It occurs when the estimated heat generation within the
29:25
system exceeds the system's ability to dissipate that heat.
29:29
So heat in is greater than heat out.
29:31
That's it. The first step in diagnosis is
29:34
classifying the heat generation sources by magnitude.
29:37
Is it the gear mesh, The bearings?
29:38
Or is it mostly churning loss? Engineers estimate heat
29:41
generation by quantifying every source of power loss, P loss,
29:44
which is the energy converted to heat, usually in watts.
29:47
And you have to balance that against what the housing can get
29:50
rid of. Correct.
29:51
If you're calculated P loss is too high relative to the thermal
29:55
dissipation capability, you must add external cooling or accept
29:59
thermal failure. And this brings us to the
30:01
critical role of the lubricant. It's not just there to reduce
30:05
friction. No, it serves A vital dual role.
30:08
Yes, it reduces wear and friction, but equally important,
30:11
it provides A crucial cooling effect by physically carrying
30:15
the heat away from the contact zone.
30:17
For high speed, high load applications, continuous spray
30:20
or jet oil lubrication is often mandatory.
30:23
And the oil itself is cooled. Often through a heat exchanger,
30:26
specifically to control the temperature rise caused by that
30:29
high frequency meshing action. If the lubricant fails to
30:33
maintain that separation and cooling, what are the primary
30:36
failure modes we see particularly micro pitting and
30:39
scuffing. We classify these as micro
30:41
failures tied directly to lubrication breakdown.
30:44
Micro pitting is a surface fatigue phenomenon on highly
30:47
loaded case hardened gear flanks.
30:49
The key contributors are local stress, surface roughness,
30:52
sliding speed and critically, the lubricants properties.
30:56
It looks like a Gray matte surface.
30:59
Exactly. But here's an important nuance.
31:01
A certain limited level of micro pitting is sometimes
31:05
permissible. It can actually lead to a
31:07
slightly improved, more stable flank condition after a break in
31:11
period. It's when it becomes progressive
31:13
that it's a serious failure. And scuffing That sounds much
31:16
more violent. It is scuffing is a severe
31:19
failure mode often seen in highly loaded hypoid gears where
31:23
sliding velocities are maximum, like a vehicle differential.
31:26
It's caused by a brutal combination of high sliding
31:29
velocities and high contact stresses which leads to
31:32
instantaneous welding and tearing of the contact surfaces.
31:35
The metal is ripped apart. And lubricants can help prevent
31:38
that. Lubricants and their extreme
31:40
pressure, or EP additives, have a profound influence on scuffing
31:44
resistance. That seems like an area where
31:46
surface finishing technology could provide a massive safety
31:49
net. Let's talk about isotropic super
31:51
finishing, or ISF. ISF is a crucial chemical
31:54
mechanical finishing process that dramatically improves
31:58
surface integrity beyond standard grinding for scuffing.
32:01
Resistance tests have demonstrated that isotropic
32:04
super finished SAE 9310 alloy specimens show at least a 40°F
32:10
higher lubricant temperature threshold before scuffing
32:13
occurs. Compared to standard ground
32:15
gears, a 40° buffer is enormous in thermal management terms.
32:19
It is a huge safety factor. It essentially extends the life
32:22
of the lubricant. Furthermore, this technology has
32:25
been shown to increase a gears resistance to contact fatigue
32:28
pitting by a factor of 3. Then bending, fatigue, bending.
32:32
Fatigue resistance by at least 10%.
32:34
The result is aerospace grade performance, allowing gears to
32:37
operate near the upper bounds of their theoretical design limits
32:40
for far longer. That's a compelling case for a
32:42
high quality finishing operation.
32:44
Speaking of quality, let's wrap up that looking at quality
32:47
control manufacturing methods and how engineers determine why
32:50
a gear ultimately broke. The final analysis.
32:53
When designing a new system, the engineer has to specify the
32:56
required precision, what AGMA quality grade do I actually need
33:00
and how tight to gear tolerances really need to be.
33:02
Gear quality classes are standardized globally by
33:05
organizations like a GMA, ISO, DIN, and GIS, and it's vital to
33:10
understand the different scales. They're not all the same.
33:13
Not at all. AGMA uses grades eight through
33:15
15 where a higher number of means better gear tolerance.
33:19
Conversely, ISO DIN and JIS use a system where a lower class
33:23
number means better quality. So an ISO grade 3 is airspace
33:27
precision. And the need for tight
33:28
tolerances directly correlates with noise and vibration.
33:31
At load capacity, inaccurate teeth and poor finish lead
33:34
directly to those high dynamic loads we discussed earlier.
33:37
Typical automotive transmissions often require a GMA10 to 12A.
33:41
Precision robot arm might require a GMA14 or 15.
33:45
The benefit of investing in tolerance analysis is clear.
33:48
Reduce reject rates and fewer assembly problems.
33:50
But inspection is tricky. How does the type of inspection
33:53
data composite versus elemental effect performance judgments?
33:56
This is a subtle but important point.
33:58
Most inspection data provided by equipment today are composite
34:02
deviations measurements like total accumulated pitch or run
34:06
out. These comosite measurements can
34:08
often obscure the true nature of individual elemental deviations.
34:12
Like an error in the involute rofile itself.
34:15
Exactly an error in the involute, the Helix lead or the
34:18
pitch. You might have a gear that
34:20
passes the composite measurement test, but contains A localized
34:23
sharp error that will cause immense dynamic loads and noise
34:27
at that single point of contact. So composite data can be
34:30
misleading. It can lead to inappropriate
34:32
judgments, either rejecting satisfactory gears or, much
34:36
worse, passing unsatisfactory ones that fail rapidly in
34:39
service. Let's discuss the manufacturing
34:41
process itself. How does the manufacturing
34:43
method affect strength and noise?
34:46
The method dictates the geometric possibilities and the
34:48
structural integrity, particularly at the highly
34:51
stressed tooth root. Hobbing is still the versatile
34:53
workhorse for spur and helical gears, but for internal gears,
34:57
which are essential in planetary stages, power skewing is
35:00
increasingly recognized for providing high precision and
35:02
high production rates. And for the really complex
35:05
stuff. For ultra precise or previously
35:08
non technological geometries, engineers leverage 5 axis CNC
35:12
milling paired with tooth contact analysis or TCA.
35:16
This allows for the creation of unique tooth lines like cosine
35:19
shaped tooth lines which optimize tooling and more
35:23
importantly reduce edge loading under deflection.
35:26
I assume the way the gear is cut also effects its ability to
35:28
resist fatigue. Absolutely.
35:30
The roof filet area is the critical zone for strength.
35:33
That's where bending failure almost always originates.
35:36
The cutting tool geometry and process parameters greatly
35:39
influence the stress concentration factors in this
35:41
area. Optimizing that filet form can
35:44
result in significantly higher bending fatigue strength.
35:46
So after all the design work, the material selection, the heat
35:49
treatment, and the manufacturing, something still
35:51
breaks. The last line of defense is
35:53
failure analysis. Why did this gear tooth break
35:56
and how do I tell overload from fatigue failure?
35:59
Failure analysis relies on classifying the damage based on
36:02
standardized terminology. The definitive differentiation
36:05
rests entirely on the cause and the resulting fracture
36:08
appearance. It's forensic work.
36:09
So what's the key difference? Fatigue failure is caused by
36:12
repeated stress cycles that exceed the endurance limit of
36:15
the material over time. It's a gradual, progressive
36:18
failure. And it looks like.
36:20
It manifests as pitting, spalling, case crushing and
36:23
ultimately fatigue breakage. A crack that originates and
36:27
grows slowly, usually from the tooth root fillet, and it shows.
36:30
A smooth, progressive crack line, often with distinct beach
36:34
marks. So fatigue is gradual?
36:36
What about overload? That sounds sudden.
36:39
It is overload failure is caused by a sudden, unexpected heavy
36:42
single event load, a system jam, a massive torque spike.
36:46
This failure is characterized by overload, breakage or tooth
36:49
shearing, where the tooth is sheared instantly from the body.
36:52
And the fracture surface looks different.
36:54
Completely different in overload failure.
36:56
It's typically jagged, rough and crystalline, often displaying a
36:59
45° sheer lip rather than the smooth, progressive line you see
37:03
in fatigue. And what are some of the other
37:05
key failure modes we might encounter that don't fall neatly
37:08
into those two major categories? We frequently see scoring which
37:12
appears as linear scratches in the slipping direction, often
37:15
associated with lubricant film failure.
37:19
There was also fretting corrosion, which is surface
37:22
damage caused by minute reciprocal motion often seen on
37:24
splines, and finally plastic deformation, where the material
37:29
permanently flows without fracturing, leading to issues
37:32
like rolling or ridging around the pitch line.
37:34
That's a phenomenal synthesis. We covered everything from the
37:37
foundational math behind the Lewis equation and why it breaks
37:40
down at high speed, to the strategic use of flank
37:43
modifications to quiet the machine.
37:45
We did. We also discussed the critical
37:47
role of the quench process in heat treatment and why seemingly
37:51
minor factors like bearing stiffness affect load sharing in
37:55
a planetary stage. It really underscores that
37:57
robust gear design is about managing a complex,
38:00
interconnected system where material science, geometry, and
38:04
dynamics all have to be balanced perfectly.
38:06
They do. Before we wrap up, let's leave
38:08
you, the listener, with one final provocative thought to
38:11
consider that builds on all the complexity we've just unpacked.
38:14
Based on research into gear action and longevity, here is
38:17
something to Mull over. We know that running identical
38:20
gear sets, same loads, speeds and lubrication, but operating
38:24
them as speed reducers versus speed increasers can yield
38:28
strikingly different propensities for wear and micro
38:31
pitting over their lifespan. So the direction of power flow
38:34
matters. The lifespan of the system seems
38:36
quite sensitive to it. That's fascinating.
38:38
You're talking about identical physics, just run in reverse.
38:41
What fundamental differences in gear action in the sliding
38:46
velocity profiles? Maybe, or the entry and exit
38:48
points of contact might explain the sensitivity.
38:52
Why does acting as a break wear the teeth differently than
38:54
acting as a torque amplifier? That's something to research and
38:58
think about on your own time. Indeed, the pursuit of
39:00
optimization is never truly finished.
39:02
That's it for today Design junkies.
39:04
If you're getting something out of this, follow podcast, leave a
39:06
review and don't forget to check out our YouTube.
39:08
It helps the whole family grow, learn and build Better Together.
39:12
A family that solves together stays together.