Latest / Mechanical Engineering Made Simple / (#109) Solving Thermocouple Drift and Phantom Pressure

Transcript

(#109) Solving Thermocouple Drift and Phantom Pressure · Mechanical Engineering Made Simple

0:00
Welcome back to mechanical engineering made simple.
0:01
It is great to be back. Today we are walking into a
0:06
situation that I think every single one of you out there has
0:09
faced in some form or another. Oh, absolutely.
0:12
We. Have a bit of a mystery on our
0:13
hands? It's a tech thriller honestly,
0:15
but instead of international espionage and, you know, spies
0:19
and microfilms, it is set inside the high stakes, high stress,
0:25
caffeine fueled world of a thermal testing lab.
0:29
A narrative involving one of the most notoriously misunderstood
0:32
concepts in the entire engineering Canon.
0:35
Right, We are talking about the Ghost in the Machine.
0:38
Exactly. So before we get into the heavy
0:40
theory, I really want to set the scene for everyone listening.
0:42
Paint the picture. I want you to picture a specific
0:45
professional. Let us just call him the
0:46
Engineer. The engineer OK.
0:48
And he isn't a rookie. He is smart.
0:51
He is capable. He has, I don't know, 10 or
0:54
maybe 15 years of experience under his belt.
0:57
He has seen it all. Right, he is the guy you call
0:59
when the simulation just refuses to match reality.
1:03
Exactly. He knows his way around APID
1:05
Controller, a soldering iron, and he probably knows the
1:09
specific smell of burning capacitors by heart.
1:11
Oh, we all know that smell. We do, and right now the
1:15
engineer is running a critical validation test on a brand new
1:18
high efficiency engine prototype.
1:20
High stakes. The stakes are massive.
1:23
This is not just a minor tweak. This engine is supposed to be
1:26
the next big leap in thermal efficiency.
1:28
So we are talking about shaving percentages off fuel
1:32
consumption. That could mean millions of
1:35
dollars. Millions.
1:36
So it is high pressure, high temperatures and extremely high
1:40
exectations from management. Who are probably pacing outside
1:43
the glass window of the test cell right now.
1:45
Oh, without a doubt, the trifecta of engineering stress.
1:48
So the engineer sets up his test rig.
1:50
Meticulously, I assume. Very meticulously he has got Ted
1:53
K thermocouples plastered all over the exhaust manifold.
1:57
He has high sensitivity pressure sensors monitoring the flow
2:00
across a porous ceramic particulate filter.
2:02
OK, standard setup. He double checks the wiring, he
2:04
checks the grounds, flips the switch, and the engine roars to
2:07
life. The floor vibrates.
2:09
The smell of hot metal fills the air and he sits back, grabs his
2:13
coffee and looks at the data acquisition screen.
2:16
And that, inevitably, is when the trouble starts.
2:20
The numbers are lying to him. Or at least they don't make any
2:23
sense based on what he learned in his standard thermodynamics
2:26
class. Right, the temperature readings
2:29
are drifting even though the engine block thermal mass should
2:32
be totally stable by now. It is fluctuating in a way that
2:35
suggests the engine is breathing, but the RPM is
2:38
totally constant. Yes, and weirder still, his
2:42
pressure sensors on that filter are showing a gradient, A
2:45
distinct pressure difference. Even when the gas flow is
2:48
completely shut off. Even when the flow is 0.
2:51
He is seeing ghosts. He really is.
2:53
He is seeing pressure where there should be equilibrium.
2:56
He is seeing heat fluctuations where there should be steady
2:58
numbers. Oh, he starts sweating.
3:00
He checks the sensors, they seem fine.
3:02
He checks the DAQ board, it is fine.
3:05
He starts wondering if the prototype itself is failing.
3:07
Like is a seal broken? Is combustion completely
3:10
unstable? The engineer is panicking
3:12
because he thinks the machine is breaking, but the machine is not
3:16
breaking. The engine is likely perfectly
3:18
fine. What is failing him is his
3:21
understanding of the physics of measurement.
3:23
Because he was falling into a trap.
3:25
He is falling into the trap of thermostatics.
3:27
Thermostatics. OK, break that down.
3:29
Well, he is trying to apply the comfortable, simple rules of
3:33
classical equilibrium to a system that is very much alive
3:38
with flow. He is operating in the wild,
3:42
untamed world of irreversible thermodynamics.
3:45
That name always sounds so ominous to me.
3:48
Irreversible thermodynamics. It sounds like a mistake you
3:51
cannot undo, like accidentally deleting the entire production
3:54
database. In a way, it really is about
3:57
things that cannot be undone. It is the study of systems that
4:00
are not settled. Classical thermodynamics, the
4:02
stuff we all learn in Thermo One O 1 is mostly about states of
4:06
rest. You have a box of gas.
4:07
You wait a million years. What does it look like?
4:09
That is equilibrium. Right, everything has stopped
4:11
moving. Exactly.
4:12
But the engineer isn't dealing with a box of gas at rest.
4:15
He is dealing with a roaring engine.
4:18
He is dealing with systems where heat, electricity and matter are
4:22
all moving simultaneously. And more importantly, and this
4:25
is the key to this entire deep dive, they are interfering with
4:29
each other. They are crossing lanes.
4:30
Yes, to solve the engineers problem and to fix his data so
4:35
he doesn't scrap a perfectly good engine.
4:38
We really have to look at what happens when these fundamental
4:41
forces collide. And to guide us through this
4:43
ghostly terrain, we are diving deep into some source material
4:48
specifically covering irreversible thermodynamics.
4:51
We are going to focus on the heavy hitters today.
4:53
Entropy production. On Sager's reciprocal relations.
4:57
And the hidden complexities of thermoelectricity.
4:59
We are going to troubleshoot the engineer's rig step by step,
5:03
using the text to explain exactly why his sensors are
5:06
acting haunted. So let us go to the scene of the
5:09
crime, Part 1, the setup and the steady state fallacy.
5:12
OK, set it up. The engineer is looking at a
5:15
specific component, let's say it is a metal support bar
5:18
connecting the hot engine block to the cooler mounting frame.
5:22
He looks at his thermal camera, sees the gradient is completely
5:25
stable and he thinks to himself, OK, the temperatures are
5:28
constant at my sensor points, so the system is in equilibrium.
5:32
And that is his first fundamental mistake.
5:34
Why he is confusing steady state with equilibrium?
5:38
I feel like we use those terms interchangeably in casual
5:41
conversation all the time though.
5:42
Like, oh, things are settled down, we're in equilibrium.
5:45
We do, but it is a massive difference in physics.
5:48
What is the actual rigorous difference?
5:50
In classical thermodynamics, which again we should probably
5:54
call thermostatics, equilibrium essentially means death.
5:58
Thermally speaking, yes, it means everything is totally
6:01
uniform. If that bar were truly in
6:03
equilibrium, the temperature would be exactly the same at the
6:06
engine end, the middle, and the frame end.
6:08
There would be no flow of heat. Zero.
6:10
No energy moving. It would be dead.
6:12
The entropy would be absolutely maximized and nothing
6:15
interesting would ever happen again.
6:17
But in this case, one end of the bar is bolted to a combustion
6:20
engine running at hundreds of degrees.
6:22
And the other is bolted to a cold frame in an air conditioned
6:26
room. So you have a massive potential
6:28
difference. Exactly.
6:30
Heat is pouring into the hot end and draining out of the cold.
6:33
End OK, so it is flowing. Right now, if the rate of heat
6:37
entering exactly equals the rate of heat leaving, the temperature
6:42
at any specific physical point, say exactly 2 inches from the
6:45
engine block, will not change over time.
6:48
It stays at say 150° permanently.
6:51
Yes, that is a steady state. So the variables are constant in
6:54
time. But they vary in position and
6:56
importantly, there is a flux there is to flow.
6:59
It is dynamic stability, not static stability.
7:02
So to the engineer looking at a single thermometer value, it
7:05
looks calm. It looks like a still pond.
7:08
But underneath, there's a rushing river of energy.
7:11
Precisely, And that river implies something profound about
7:14
entropy. This is where the source
7:16
material gets really interesting, and where the
7:18
engineer's intuition usually fails them.
7:20
Well, I think most people, myself included back in the day,
7:23
assume that if the temperature isn't changing, the entropy of
7:26
the system isn't changing. Right, if the state is steady,
7:29
the properties are steady. Right, and they would be
7:31
completely wrong. Tell me why.
7:33
The source material makes a very specific, very sharp point about
7:36
this. Even in a steady state RAW that
7:39
looks completely static to the naked eye, entropy is being
7:43
actively produced. Manufactured.
7:45
Yes, it is being generated constantly.
7:48
But how? If the state isn't changing,
7:50
where is the entropy coming from?
7:53
Usually we associate entropy with things physically getting
7:56
messier or changing state. Think about the reservoirs, you
7:59
have a hot reservoir at a temperature we will call T1 and
8:03
a cold reservoir at T2, OK. T1 is hot, T2 is cold.
8:07
Heat flows from hot to cold. That is just the second law of
8:09
thermodynamics. OK, now follow the entropy
8:12
calling it. When a tiny packet of heat,
8:14
let's call it DQ, leaves the hot reservoir, that hot reservoir
8:18
loses some entropy. Right, because the definition of
8:21
entropy changes. Heat divided by temperature
8:22
change in South equals DQ divided by T.
8:25
Exactly. So the hot side loses an amount
8:28
of entropy equal to DQ divided by a large number T hot.
8:32
Because the temperature is high. But here is the catch.
8:36
The cold reservoir gains that exact same amount of heat
8:39
energy. Energy is conserved.
8:41
Right DQ goes into the cold side.
8:43
But because its temperature is lower, the entropy it gains is
8:46
actually larger. Wait, run that by me again.
8:49
Why is the entropy larger if the energy is the same?
8:51
Because of the denominator in the math, you are dividing the
8:54
exact same amount of energy DQ by a smaller number T cold.
9:00
When you divide by a smaller number, you get a bigger result.
9:02
Like the money analogy? Exactly, if I give you $10.10
9:07
dollars matters a lot more to you if you are completely broke
9:10
than if you are a billionaire, right?
9:12
The cold reservoir values that heat much more in terms of
9:15
entropy. So the entropy gained by the
9:17
cold side is mathematically larger than the entropy lost by
9:20
the hot side, correct? So the net result?
9:22
The net result is positive. The total entropy of the
9:25
universe has increased. The time derivative of S which
9:28
we can call DS divided by IDT is strictly greater than 0.
9:33
That is the production. Yes, the source text describes
9:36
this as entropy production. Even though the rod looks
9:40
entirely static, it is a factory generating entropy.
9:44
It is taking high quality energy hot heat and degrading it into
9:48
low quality energy cold heat. And this production isn't just
9:52
some philosophical footnote for textbooks.
9:54
This production is the actual engine driving the hidden forces
9:58
that are messing up the engineer's measurements right
10:00
now. He thinks he's looking at a
10:02
still pond. But he is standing in a current
10:04
that is generating massive disorder.
10:06
That is a fantastic image. He is standing in a river of
10:10
disorder, and that current doesn't just stay in its lane.
10:13
No it does not. Which brings us to Part 2, the
10:16
first glitch. The glitch.
10:18
The engineer notices that the heat flow isn't just moving
10:21
heat, it seems to be messing with his electrical readings and
10:24
even the gas diffusion in that particular filter.
10:27
This is where we have to introduce the core concept of
10:29
coupling. This is the absolute heart of
10:32
irreversible thermodynamics. Usually in our undergraduate
10:35
classes we have these really nice separate laws.
10:38
Clean equations. Right, we have Fourier's law,
10:40
which says temperature difference drives heat flow.
10:43
Simple. Then we turn the page.
10:45
To Ohm's Law. Exactly Ohm's law which says
10:48
voltage difference drives current.
10:51
Then maybe fix law a few chapters later where a
10:54
concentration difference drives diffusion.
10:56
They stay in their separate chapters.
10:58
Right, They're polite neighbors who stay in their own yards.
11:00
But the source material points out that in the real world, in
11:04
the messy, highly energetic world of engine testing, these
11:08
forces interfere. They cross couple.
11:11
Meaning heat can drive electricity.
11:13
And electricity can drive heat. And heat can drive mass
11:16
transfer. And mass transfer can drive
11:17
temperature changes. It is a massive interconnected
11:20
web. This sounds like absolute chaos
11:22
for an engineer. If everything effects everything
11:25
else, how do we even model that? How do you write an equation for
11:27
everything is happening at once? We use a brilliant framework
11:31
developed by Lars on Sager. He was a theoretical physicist.
11:34
OK, how's it work? The source introduces a set of
11:36
linear equations he developed. Basically you have a flow which
11:41
we will call J and it equals a coefficient l * a driving force
11:47
X. Flow equals coefficient times
11:49
force. Yeah, J equals L * X.
11:53
That sounds simple enough. That's just like Ohm's Law.
11:56
It starts out that way, but when you have coupled flows it
11:59
becomes a matrix of equations. A matrix.
12:02
Let's say you have heat flow and electrical flow happening in the
12:05
exact same wire. The equation for heat flow
12:08
doesn't just depend on the temperature difference anymore.
12:10
It has a term for the voltage difference too, yes.
12:13
So heat flow equals coefficient 1.
12:15
So one times temperature difference plus.
12:17
Plus coefficient 1/2 times the voltage difference.
12:21
That second term with the coefficient L sub 1/2, that is
12:24
the cross coupling term. Exactly.
12:26
It represents the interference. It tells you exactly how much
12:29
the voltage is physically pushing the heat around.
12:31
And this is where on Sager won his Nobel Prize.
12:34
It wasn't just for writing down a matrix.
12:35
Anyone can write a matrix. No, writing the matrix is just
12:38
algebra. His Nobel Prize and the core
12:41
insight of this entire source material is the discovery of the
12:45
reciprocal relations. Reciprocal relations.
12:48
What did that mean? He proved that the matrix is
12:50
symmetric. That means the cross coupling
12:53
coefficient L sub 1/2 is exactly mathematically equal to the
12:59
coefficient L sub 21. Pause on that.
13:01
That sounds a bit abstract. What does that mean in plain
13:03
English for our engineer in the lab?
13:05
It means nature has a deep fundamental symmetry.
13:09
It means that the effect of temperature on electricity is
13:12
mathematically identical to the effect of electricity on heat.
13:16
So if a temperature difference can cause a voltage.
13:18
Which we call the C Beck effect. Then an electric current must
13:21
carry heat. Which you call the Peltier
13:22
effect. You literally cannot have one
13:24
without the other. There are two sides of the same
13:26
thermodynamic coin. Exactly.
13:28
You cannot design A sensor that uses the C Beck effect without
13:32
inviting the Peltier effect to the party.
13:34
The math forbids it. So why does this matter to the
13:37
engineer? Right now he is just trying to
13:39
read a simple temperature on the screen.
13:41
Because it validates his confusion.
13:44
He isn't crazy. He isn't imagining things.
13:47
The physics of the universe demands that these things
13:49
overlap, right? If he has a temperature gradient
13:52
on his test rig, he has electrical potential gradients
13:55
happening whether he wants them there or not.
13:57
He is dealing with phantom voltages.
13:59
And phantom heat flows. Which leads us directly to his
14:03
biggest headache of the day. Part 3.
14:06
Troubleshooting the thermocouple.
14:07
The bread and butter of thermal testing.
14:09
But evidently it is vastly more complex than just twisting 2
14:13
dissimilar wires together and reading a dial.
14:16
It is the classic temperature problem in engineering.
14:19
So let's look at what the engineer is doing.
14:21
He is using a standard thermocouple setup.
14:23
He has his two wires. He has a Volt meter, which the
14:27
source refers to as a potentiometer to be precise,
14:29
right? And he's trying to measure the
14:31
exhaust temperature, right? But the reading is off, it is
14:34
drifting around. It does not match his
14:36
calibration data at all. To understand why it is
14:39
drifting, we really have to look at what a thermocouple actually
14:42
measures. OK, most people think it
14:44
measures temperature exactly at the tip where the wires meet.
14:47
It doesn't. I mean I thought the little
14:48
welded bead at the end was the sensor.
14:51
That is the most common misconception.
14:53
The source material goes through a very rigorous derivation
14:56
involving a concept called the entropy of transport.
14:59
Entropy of transport. They denote that as *.
15:01
Right. Yes, Star, and it is the
15:04
absolute key to the whole mystery.
15:06
What is it? The entropy of transport is
15:08
basically the rate at which entropy is carried along by the
15:12
flow of electrons. Okay, imagine the electrons as
15:15
little trucks on a highway. Perfect analogy.
15:18
They aren't just carrying electrical charge in their cargo
15:20
beds, they are also carrying thermal energy and disorder.
15:24
Entropy. Right when electrons move from a
15:26
hot zone to a cold zone, they drag this entropy with them down
15:30
the wire. So the electrons are
15:31
multitasking. Very much so now looking at the
15:34
sources analysis of the measurement circuit, we have
15:37
wire X and wire. Y Those are the two different
15:40
metals of the thermocouple, say Chromel and Alamel.
15:43
Yes. Then we have the leads, let's
15:46
call them lead Z, which are usually just standard copper
15:49
wires connecting back to the voltmeter.
15:51
OK, the extension wires. The total voltage that the meter
15:54
actually reads is the sum of the potential changes across all of
15:58
these wire segments combined. And the source integrates this S
16:02
star value across the temperature differences of each
16:05
individual wire segment. Correct.
16:07
The mathematical derivation shows that the measured voltage
16:10
depends entirely on the difference in S star between the
16:13
two metals integrated from the cold temperature to the hot
16:16
temperature. So it is an integral of the
16:18
quantity S star for metal A -, S star for metal B with respect to
16:24
temperature, yes. With respect to DT.
16:26
So far so good. Yeah, that is how a thermocouple
16:29
fundamentally works. The difference in how the two
16:31
different metals carry entropy creates the voltage we read as a
16:35
temperature. Right?
16:36
So where is his error coming from?
16:38
Why the drift? The error comes from the
16:40
connection the plugs. Yes, the source highlights a
16:42
very specific trap in the layout, often labeled as point
16:46
C&E in the circuit diagram. What are those points exactly?
16:49
These are the physical points where the thermocouple wires
16:52
actually connect to the copper lids that go to the the meter.
16:56
The junction between the specialized sensor wire and the
16:59
cheap extension. Wire exactly, usually inside a
17:03
plastic connector plug or on a screw terminal strip on the data
17:07
acquisition board. OK, I kicked you that.
17:09
The golden rule, the absolute requirement derived directly
17:12
from that S star interval, is that those two connection points
17:15
must be at the exact same temperature.
17:17
Why? What happens if they aren't?
17:20
Because if they aren't at the same temperature, they act as a
17:22
secondary parasitic thermocouple.
17:25
Think about it. The leads are likely copper
17:28
material Z if connection Point C is sitting at 30°C and
17:33
connection point E is at 35°C. You have created a new
17:37
thermocouple junction between the sensor wire and the copper.
17:40
Exactly. And that parasitic junction
17:42
generates its own tiny voltage. Which simply adds or subtracts
17:46
from your main signal. Right, the potential generated
17:48
at Point C doesn't cancel out the potential generated at point
17:51
D because the temperatures aren't identical.
17:54
And the engineer mentioned earlier that his engine
17:56
prototype is huge. It's a massive piece of metal.
17:59
There are thermal gradients everywhere in that test lab.
18:01
Precisely so. Picture his exact setup on the
18:05
bench. He probably has his DAQ box
18:07
sitting on a folding table near the engine.
18:10
Sure, maybe the tiny cooling fan from his laptop is blowing
18:13
slightly on one side of the terminal block and not the
18:15
other. Or maybe one wire is just pulled
18:17
a bit tighter and sits closer to the radiating heat of the engine
18:21
block. Exactly that small, maybe 5°
18:24
temperature difference across those two screw terminals is
18:27
quietly injecting a rogue voltage into his measurement
18:31
circuit. So his quote UN quote hot
18:33
junction reading on the screen is actually his hot junction
18:36
voltage minus this parasitic junction voltage.
18:39
Yes, he thinks the engine exhaust temperature is drifting
18:42
around, but actually the room air temperature near his plastic
18:46
connector is drifting. That is madding.
18:48
It happens all the time. How does he fix it once he
18:51
realizes this is irreversible? Thermodynamics at play?
18:55
What is the practical fix? The narrative fix is
18:58
surprisingly simple once you understand the underlying
19:00
thermodynamics. He needs to isothermalize those
19:03
connections. Isothermalize, meaning make them
19:06
the same temperature. Right, he needs to physically
19:08
clamp the connection points together, electrically insulated
19:11
from each other obviously, so they don't short out.
19:13
But thermally coupled. Yes, so they're forced to be at
19:16
the exact same temperature. Like sinking them into a solid
19:19
block of metal. Ideally, yes.
19:22
An isothermal block. A thick chunk of aluminum or
19:25
copper that holds both screw terminals securely.
19:28
Because the metal block acts as a thermal bridge balancing them
19:32
out. Exactly.
19:33
That ensures that temperature T sub C perfectly equals
19:36
temperature T sub E. And once he does that, what
19:40
happens to the math? The potentials at those
19:42
connection points perfectly cancel each other out because
19:45
the integral from T sub C to T sub B is now an integral from,
19:48
say 30° to 30°. Which is 0.
19:51
Exactly. And the reading on his screen
19:54
immediately stabilizes. He is finally seeing the true
19:57
temperature of the exhaust. The first ghost is gone.
20:00
OK, so he fixes the drift. He is feeling a lot better.
20:05
He takes a sip of his cold coffee.
20:06
Small victories. But the engineer is thorough.
20:10
He has been burned before, so he is paranoid now.
20:13
Good engineer. He wants to check the physical
20:15
resistant of his sensor wires just to be absolutely sure they
20:18
aren't damaged or corroded from the heat.
20:20
The logical next step? So he switches his multimeter
20:24
over to the ohm setting and injects a small electrical
20:27
current into the circuit to measure the resistance.
20:30
And that innocent action triggers the second glitch.
20:33
Part 4 The current and the heat. Here we go.
20:36
As soon as he runs that tiny diagnostic current, the
20:39
temperature profile of the wire physically changes.
20:42
The steady state he just achieved is totally ruined.
20:45
This is where the cross coupling we talked about earlier, those
20:47
on savior relations, hits him right in the face.
20:50
Why? What is he assuming will happen?
20:52
He assumes that running a current just creates standard
20:55
resistance heating. You know I squared R losses
20:58
Jewel heating. That is standard electrical
21:00
theory. Wires get warm when current
21:03
flows through them. Every engineer knows that
21:05
intuitively. But in irreversible
21:07
thermodynamics there is more going on.
21:09
There is the Peltier effect. We mentioned that briefly, the
21:13
exact reciprocal of the Seebeck effect.
21:15
Yes, the source defines the Peltier coefficient in direct
21:19
relation to the difference in the entropies of transport
21:22
between the two metals. So going back to S *.
21:24
Exactly. When an electrical current flows
21:27
across the physical junction of metal X and metal Y, it doesn't
21:31
just pass through seamlessly. What does it do?
21:34
It must either absorb or release thermal heat to satisfy the
21:38
fundamental entropy balance. So unlike standard resistance
21:41
heating, which always just heats up the wire regardless of which
21:44
direction the current is flowing.
21:45
Peltia heating is fully reversible.
21:47
It is directional. Directional heat.
21:49
Yes, if the current flows one way, the junction heats up.
21:53
If you reverse the battery and the current flows the other way,
21:55
the junction actively cools down.
21:57
It acts like a tiny solid-state refrigerator.
22:00
Exactly. It pumps heat.
22:01
So by simply checking the resistance with his meter, the
22:04
engineer is actually turning the very tip of his sensor into a
22:07
microscopic refrigerator or heater.
22:10
He is physically pumping heat into or out of the exact
22:14
measurement point. He is wildly distorting the very
22:18
physical property he is trying to measure.
22:20
He is changing the temperature of the probe tip just by asking
22:23
it a question. That is exactly what is
22:25
happening. That is incredible.
22:27
But there is another effect at play here too, isn't there?
22:29
One that is even more subtle than the Peltier effect.
22:32
Ah yes, the Thompson effect. The deep cut.
22:35
That is the deep cut of irreversible thermodynamics.
22:39
The Peltier effect only happens at the distinct junction where
22:42
two different metals meet right? But the Thompson effect happens
22:46
along the entire continuous length of a single uniform wire.
22:49
Provided there is a temperature gradient.
22:51
Exactly. Not at a junction, but in the
22:53
bulk wire itself. Walk us through the mechanism of
22:56
that. Imagine the electrons flowing up
22:58
the temperature gradient. They are moving from the cold
23:01
end of the wire outside the engine up to the hot end
23:04
attached to the exhaust. OK, think of those electrons
23:07
like hikers climbing a very steep mountain.
23:10
I like this analogy. The electrons are the hikers,
23:13
the temperature gradient is the physical mountain.
23:15
Right now, to climb that mountain, to move into the much
23:19
hotter, highly energetic region of the metal, the hikers have to
23:23
physically gain energy. They need to match the higher
23:26
thermal energy of the vibrating metal lattice around them as
23:30
they go up. Precisely.
23:31
So where do they get that extra energy?
23:33
They have to take it from the mountain itself.
23:34
Exactly. They actively absorb heat from
23:38
the wire lattice itself to fuel their thermal climb.
23:41
And by absorbing that. Heat.
23:42
They locally cool the wire down as they pass through.
23:45
Wow. And what if they flow downhill?
23:48
If the current reverses and they flow from hot to cold, they are
23:52
descending the mountain. So they have too much energy.
23:54
Right, they have excess thermal energy they need to shed in
23:57
order to match the cooler, calmer surroundings at the
24:00
bottom. So they release heat into the
24:01
wire. Exactly.
24:03
The source derives the mathematical calculation for
24:05
this. The Thompson heat is strictly
24:07
proportional to three things, the electrical current, the
24:11
temperature gradient, and the specific Thompson coefficient of
24:14
the material. So let's look at the engineer's
24:16
setup. He has a diagnostic current
24:19
flowing through a sensor wire that runs from a 500° exhaust
24:23
manifold out to a 20° room. That is a massive thermal
24:28
gradient. Huge.
24:29
And as that meter current flows, it is actively redistributing
24:33
heat all along the entire length of that wire.
24:37
It is physically warping the temperature profile of the
24:39
metal. It is not just heating up
24:41
uniformly like a cheap toaster element, it is scooping up heat
24:44
from one part of the wire and dumping it into another part.
24:47
The engineer suddenly realizes that by measuring the
24:50
resistance, he is actively changing the state of the
24:53
system. This is effectively the
24:55
Heisenberg uncertainty principle of macro thermodynamics.
24:58
You literally cannot measure the system without disturbing it.
25:01
Especially if he's current. Current is the disturber.
25:03
So what is his solution? How does he check his sensor or
25:06
read the data without breaking the physics?
25:08
He must measure the voltage at 0 current. 0 current.
25:12
He needs a high impedance voltmeter that draws virtually
25:15
no electrons to take its reading.
25:17
The source heavily emphasizes that the pure C Beck voltage is
25:22
strictly A0 current phenomenon. Exactly.
25:24
If J electrical the flow of electrons is not zero, you are
25:28
immediately inducing Peltier and Thompson distortions on top of
25:32
your signal. So he cannot leave his meter on
25:35
the resistance setting or use any DAQ channel that draws
25:38
significant current. No, he has to trust the purely
25:42
passive voltage reading. So he flips the switch back.
25:45
He shuts off the current injection.
25:46
He uses a high impedance channel.
25:48
And the drift vanishes. The temperature readings are now
25:50
rock solid. He wipes the sweat off his brow.
25:53
He is feeling like a genius. But the test is not over.
25:56
No it is not, because he still has those pressure sensors on
26:00
the porous filter. Filter.
26:02
Part 5 The third glitch, and this is the one that really
26:05
freaks him out. It would freak anyone out.
26:07
So he has this highly specialized ceramic filter
26:10
sitting in the exhaust line. The engine is off now, but the
26:14
heavy metal exhaust pipe is still incredibly hot.
26:16
It holds a lot of thermal mass. Right.
26:19
And there's absolutely no gas being pumped through the engine.
26:21
The mass flow rate is identically 0.
26:23
OK, zero flow. But the pressure gauge mounted
26:26
across that porous filter shows a distinct measurable pressure
26:31
difference building up. A pressure head building up with
26:34
absolutely no pump running. He thinks it is a leak in the
26:37
test cell or a fundamentally broken sensor.
26:41
He is ready to tear the whole rig apart.
26:43
He is sitting there thinking, did I leave a bypass valve open
26:47
somewhere? Is the sensor 0 point completely
26:51
drifting out of spec? But it isn't any of those
26:53
things. No, it is actually a really
26:56
beautiful real world demonstration of coupled
26:58
diffusion. Coupled.
26:59
Diffusion. Specifically, the deep coupling
27:02
of mass transfer and heat transfer.
27:04
The source text talks about the sword effect and the Dufor
27:07
effect right around here. Right, the sword effect is when
27:10
a strong temperature gradient causes a concentration gradient
27:13
to form. Like separating a mixed gas
27:15
simply by heating one side of the container.
27:17
Exactly, and the dufor effect is the exact reverse of that.
27:20
But in the specific context of the engineer's or a ceramic
27:23
filter, we're looking at something that text calls
27:26
thermal transpiration. Also known as Knudsen flow.
27:29
Knudsen flow. That sounds like something
27:31
straight out of aerospace or rocket science.
27:34
What exactly is happening inside those tiny ceramic pores?
27:38
The source explains this phenomenon using a new concept
27:41
called the heat of transport. They label that as Q star,
27:44
right? Yes.
27:44
Earlier we had S star for the entropy carried by electrons.
27:48
Now we have Q star for the energy carried by physical gas
27:51
molecules. OK, what makes this Knudsen flow
27:54
behave differently from, say, normal airflow through a duct?
27:58
It entirely depends on the physical size of the holes in
28:01
the filter. The pores.
28:03
Right. Imagine the pores in this
28:04
ceramic filter are incredibly small.
28:06
Specifically, they must be smaller than what we call the
28:09
mean free path of the gas molecules mean free.
28:12
Path explain that for us. It is the average distance of
28:15
molecule travels before it collides with another molecule.
28:18
OK, so in a big white exhaust pipe, molecules are hitting each
28:21
other constantly. Trillions of collisions.
28:23
Yes, they act like a fluid, but inside these microscopic pores
28:27
they barely ever hit each other. The pore is too narrow.
28:30
They mostly just bounce off the physical ceramic walls.
28:33
So they're essentially flying solar flow through the tunnel.
28:35
Exactly. Now let's look at the
28:37
temperatures. You have trapped hot gas on one
28:40
side of the filter and cooler room temperature gas on the
28:43
other side. Because the engine block is
28:45
still hot on one side. The molecules on the hot side
28:48
are moving much much faster. That is what temperature is.
28:51
They have higher kinetic energy. Makes total sense.
28:54
But the source points out a fascinating statistical
28:57
selection mechanism happening at the entrance to the pore.
29:01
A selection mechanism. Because the hot molecules are
29:05
moving faster and physically bouncing around much more often,
29:08
the specific ones that actually managed to randomly hit the tiny
29:12
hole and pass through to the other side carry a higher energy
29:16
than the statistical average of the hot gas.
29:18
Oh I see, they are the highly energetic outliers.
29:21
Right, they are the Hotshots of the gas world.
29:23
So these hotshot molecules zoom right through the pore to the
29:26
cold side. They do, but here is the
29:29
completely counterintuitive part.
29:31
You would normally think the pressure would just perfectly
29:34
equalize, right? High pressure pushes to low
29:37
pressure until it's flat. Always.
29:40
That's fluid dynamics one O 1. But because of this severe
29:43
temperature difference and the specific way the solar molecule
29:46
statistically hit the hole, you actually get a net flow of gas
29:50
molecules from the cold side back over to the hot side.
29:54
Wait, wait. From the cold low energy side
29:56
pushing back to the hot side that sounds completely
29:59
backwards. It does.
30:00
It's called the fountain effect. The severe temperature gradient
30:04
literally forces gas mass to pile up on the hot side of the
30:07
filter. It is creating a pressure
30:09
difference, a delta. Yes, just to balance out the
30:13
thermodynamic Dr. of the temperature difference, the
30:16
delta T. The source actually gives a
30:17
specific mathematical equation for.
30:19
This doesn't. It it does the Knudsen equation,
30:21
it rigorously relates the final pressure ratio across the filter
30:25
to the square root of the temperature ratio.
30:27
So pressure 1 divided by pressure 2 equals the square
30:31
root of temperature 1 divided by temperature 2.
30:33
Exactly. So let's look at the engineer.
30:35
He is staring at a pressure gauge that reads A1 PSI of
30:40
pressure, and he thinks there's a pump running or blockage, but
30:44
actually the pressure is there simply because the ceramic
30:48
filter is 200° hotter on the engine side than on the tailpipe
30:52
side. The temperature gradient itself
30:54
is physically creating the pressure head.
30:56
The heat is doing the mechanical work of a pump.
30:59
There is just wild. It is like the heat is a
31:02
physical wall pushing the gas backward.
31:05
It is. The engineer was frantically
31:07
looking for a leak or a broken wire, but what he actually found
31:10
was the heat of transport at work in the real world.
31:13
The gas molecules are just blindly obeying the fundamental
31:17
laws of coupled flow. They.
31:18
Have no choice. So how does he fix his data?
31:21
He doesn't need to replace the expensive pressure sensor,
31:23
right? Not at all.
31:24
The sensors working perfectly. He just needs to account for the
31:27
thermal transpiration effect in his data calibration.
31:30
If his DAQ system knows the temperatures on both sides of
31:34
the filter, which it does because he fixed his
31:36
thermocouples earlier, he can use the Knudsen equation to
31:39
mathematically calculate the false thermally induced pressure
31:43
head. And then he just simply
31:44
subtracts that calculated value from his raw sensor reading.
31:48
Exactly. He applies the thermodynamic
31:50
correction factor and suddenly his baseline pressure reads 0,
31:54
exactly as it should with the engine off.
31:57
The engineer finally sits back in his chair.
31:59
The mysterious temperature drift is entirely gone.
32:03
The phantom pressure head is perfectly explained and
32:06
corrected. The prototype engine is actually
32:08
performing perfectly within spec.
32:10
And he has officially mastered irreversible thermodynamics.
32:13
Well, he survived it at least. He didn't just fix a physical
32:17
test rig. He finally understood the deep,
32:20
invisible connections between these forces.
32:22
He realized that in the real engineering world, you literally
32:25
cannot touch one variable without wiggling three others.
32:28
It is a really good feeling when the physics finally clicks.
32:31
We have essentially saved his $1,000,000 project.
32:34
All in a day's work. But I know our listeners,
32:37
they're engineers, they're detail oriented.
32:40
They always have specific questions.
32:41
They want to know the edge cases, so let's move right into
32:44
Part 6, Frequently Asked Engineering Questions.
32:47
Let's do it. I love these practical boots on
32:50
the ground queries. Question one.
32:53
We talked about the physical wire length of the thermocouple
32:55
earlier. Does the actual length of the
32:58
thermocoule wires affect the accuracy of the reading?
33:01
Good question. Like if the engineer has to run
33:03
a massive 100m wire all the way from the test cell back to a row
33:07
load control room, does that long distance degrade the
33:11
thermal signal? It is a very common worry when
33:14
you run long wires. You naturally worry about
33:16
electrical resistance drops, you worry about picking up ambient
33:19
noise, you worry about line capacitance.
33:22
Sure. But strictly regarding the
33:23
thermodynamics of the signal generation, the definitive
33:26
answer is found right there in the S star integral we discussed
33:29
earlier. The integral of S star with
33:31
respect to temperature. Yes, the integral of S star DT.
33:35
Notice what specific variable is not anywhere in that
33:38
mathematical equation. Distance exactly?
33:41
Distance DDX or length is nowhere to be found.
33:44
The total generated voltage depends entirely and only on the
33:48
temperature difference between the physical endpoints.
33:50
So 1m wire and a 1 kilometer wire will give the exact same
33:55
millivolt output for the exact same temperature difference.
33:58
Theoretically, yes, 100%, as long as the material of the wire
34:04
is perfectly consistent, meaning is metallurgically homogeneous
34:08
throughout its entire length. You could tie the wire in
34:11
physical knots, you could loop it around the outside of the
34:14
building. It mathematically does not
34:15
matter. The voltage is what we call a
34:18
state function of the endpoint temperatures.
34:20
The path it takes doesn't matter, only the start line and
34:23
the finish line. Exactly.
34:25
That is incredibly reassuring. Unless of course, the wire is
34:28
isn't actually homogeneous. Right, and that is the real
34:31
world catch. If the long wire has bad spots
34:34
or kinks or severe rust, that physical damage changes the
34:38
local S star value of the metal. And then the position of that
34:42
damage in the temperature gradient suddenly does matter.
34:44
It matters a lot, but assuming good quality wire length is
34:48
totally irrelevant to the core physics.
34:50
Perfect. Moving to question 2, Cold
34:54
junction compensation. We hear this specific term
34:57
thrown around all the time in instrumentation manuals.
35:00
Constantly. Why do we actually need it?
35:02
Why can't the DAQ just measure the hot temp directly at the tip
35:06
and be done with it? Because of the deeply relative
35:09
nature of the measurement itself.
35:11
Look back at the circuit diagrams in the source text.
35:14
You never actually measure a discrete temperature directly
35:17
with a thermocouple. You don't.
35:18
No. You only measure a voltage
35:20
difference. And that voltage difference is
35:22
generated strictly by the physical difference in
35:25
temperatures between the two ends.
35:26
Exactly. The thermocouple only tells your
35:29
computer one thing. It says the hot end out there is
35:32
exactly 500° hotter than the cold end right here.
35:35
OK, that relative difference is the only piece of information it
35:38
gives you. If your computer does not know
35:40
what the absolute temperature of the cold end is, that relative
35:44
information is completely mathematically useless.
35:46
It is like me saying, hey, I am exactly 5 inches taller than
35:50
Bob, right? But if you have no idea how tall
35:53
Bob is, you have no idea how tall I am.
35:55
That is the perfect analogy. The cold junction is Bob.
35:58
I like that. Ideally, in a perfect lab
36:00
setting, you physically plunge the cold junction wires into an
36:04
actual ice bath at exactly 0°C. Because then Bob is at 0.
36:09
The math. Exactly that is the physical
36:11
reference we all agree on. But out in the field, like
36:14
inside the engineer's plastic control panel, the quote UN
36:17
quote cold junction is just the ambient air where the wires
36:21
physically plug into the silicon chip, right?
36:23
So the smart chip has a tiny built in thermistor that locally
36:26
measures its own physical temperature, cold junction temp,
36:30
and it mathematically adds that baseline back into the total
36:33
reading. That specific math operation is
36:36
cold junction compensation. Without it, your exhaust reading
36:39
would falsely drift up and down every single time the room air
36:42
conditioner cycled on and off. Because the baseline is
36:44
shifting. That makes total sense now.
36:47
OK, Question 3A listener asks can I measure temperature with
36:51
just one single metal wire? Why do I always need a pair of
36:55
two different metals? Good theoretical question.
36:58
This loops right back to the entropy of transport concept.
37:01
S star again. Always S star.
37:03
If you have a single uniform wire, say just pure copper wire,
37:08
and you heat one end with a torch, electrons absolutely do
37:12
physically flow. There is a gradient.
37:15
The hikers climb the mountain. Yes, but if you actually try to
37:18
measure that generated voltage, you have to attach 2 metallic
37:22
voltmeter leads to the ends of the wire.
37:25
Which are themselves also made of.
37:26
Probably standard copper. And if there are different
37:29
metals? You just accidentally built a
37:30
thermocouple, right? But if you use the exact same
37:33
pure copper metal for the entire return path to the meter, the
37:37
voltage generated by the hike was going on the way up the hot
37:40
temperature gradient. Is exactly perfectly cancelled
37:43
out by the voltage generated by the hikers on the way back down
37:46
the cooler return wire. Precisely because the inherent S
37:50
star property is identical for both paths, the thermal push up
37:54
the wire is perfectly equal to the thermal pole down the wire.
37:57
So the total voltage integral simply loops right back to 0.
38:00
Every time you physically need 2 completely different materials
38:05
with two completely different S star values so that the thermal
38:09
push going one way is fundamentally stronger than the
38:12
thermal push coming back. And that tiny net difference is
38:16
the actual millivolt signal your meter reads exactly.
38:19
However, the source text does mention one very specific
38:24
exception to this pair of metals rule.
38:26
It does the Thompson effect right.
38:28
You absolutely cannot measure the C Beck effect of a single
38:32
isolated metal, but you can physically measure the Thompson
38:35
coefficient of one single solitary material.
38:37
Why? Because that specific effect
38:39
happens continuously along the bulk length of the wire itself,
38:44
not exclusively at a physical junction between two things.
38:46
Is it an intrinsic property? It is the only thermoelectric
38:49
property that is purely intrinsic to a single material.
38:52
Measuring it is like measuring the specific physical friction
38:56
of the mountain path itself, rather than comparing two
38:59
different mountains. That is fascinating.
39:01
All right, final question from the mailbag.
39:02
And this one comes up literally whenever we start talking about
39:04
the phrase production of entropy.
39:06
I can guess what it is. It sounds highly suspicious to a
39:09
lot of engineers. The question is, is this
39:12
perpetual motion? Are we somehow getting free
39:15
energy out of this steady state system?
39:17
The answer is no. Absolutely.
39:20
Fundamentally, unequivocally no. Good to be clear.
39:23
And the source text is very, very firm on this exact point.
39:27
While these dynamic systems can settle into a mathematically
39:30
stable steady state, they are absolutely not free lunches.
39:34
Explain that simply, If the physical state is perfectly
39:37
steady and unchanging, why does it continually cost us energy?
39:41
Remember what we said at the very beginning?
39:43
Entropy is actively being produced.
39:46
The term DS divided by DT is a strictly positive number, right?
39:50
That mathematically means the entire universe is continually
39:53
becoming more and more permanently disordered because
39:55
of your rod. OK, To artificially maintain
39:58
that steady state, to forcefully keep the metal rod hot at one
40:01
end and freezing cold at the exact other end, you have to
40:04
continuously pump fresh energy into the system.
40:07
You are paying a permanent energy tax.
40:08
Yes, you are physically burning expensive diesel fuel in the
40:13
engine just to keep the hot side hot.
40:15
You are running a massive electrical cooling fan to
40:18
actively keep the cold side cold.
40:20
You are paying dearly in Watts to maintain that steep gradient.
40:24
The continuous entropy production is basically the
40:27
universe's thermodynamic bill coming due every single second.
40:30
Wow. It is definitely not perpetual
40:32
motion. It is the exact tragic opposite.
40:35
It is continuous, unavoidable energy dissipation.
40:38
And the moment you stop paying that bill?
40:40
The moment you shut the roaring engine off.
40:42
The system immediately slides back down into a dead
40:44
equilibrium and all the magical thermal flows just stop.
40:47
The ghost vanishes. Well, we have walked completely
40:50
through the solid rod, the delicate thermocouple, the stray
40:54
electrical current and the mysterious porous filter.
40:57
It really changes how you physically look at a simple
40:59
bench top test setup. It absolutely does.
41:02
It is not just inert plumbing and dumb copper wiring.
41:05
It is a highly dynamic, almost living system of interlocking
41:08
flows. We tend to get lazy and think of
41:11
engineering as just static structural blocks.
41:13
Like this over here is a wire and this over here is a pipe and
41:17
they don't talk. But irreversible thermodynamics
41:20
violently reminds us that everything is continuously
41:23
flowing and everything is deeply connected.
41:25
That is the whole point. What?
41:26
Is the absolute biggest take away for you personally from
41:29
diving into this specific text. For me, it is definitely the
41:33
profound philosophical connection between the massive
41:36
macro scale and the tiny micro scale that was so well, we spent
41:40
this entire deep dive talking about Lars on Sager's relations,
41:43
right? These big clunky macroscopic
41:46
phenomenological laws where the massive matrix coefficient L sub
41:49
one 2 = L sub 21. Right, it is a very practical
41:53
wrench turning engineering level tool.
41:55
It is, but the very end of the source material leaves us with a
41:58
truly provocative thought in its final section on statistical
42:01
mechanics. Right, the rigorous mathematical
42:03
derivation. Yes, Onsager did not just
42:06
arbitrarily guess these symmetric matrix relations
42:08
because they look pretty. He rigorously derived them from
42:11
the pure chaos of microscopic molecular motion.
42:15
He looked down at the atomic level.
42:16
He looked at exactly how trillions of microscopic
42:19
particles randomly fluctuate and then slowly regress back to
42:23
thermal equilibrium. And what did he find?
42:25
He mathematically showed that the strict reversibility of time
42:29
at the quantum atomic level. The fact that individual
42:32
bouncing atoms physically do not care which direction the arrow
42:36
of time flows. Exactly that microscopic time
42:40
symmetry is the exact physical mechanism that creates the
42:43
beautiful predictable symmetry in the giant macroscopic
42:47
engineering matrix. So the totally random chaotic
42:50
bouncing of invisible atoms. It actually strictly proves the
42:54
rigid practical laws of the engineer's thermocouple on the
42:57
bench. That is incredible.
42:59
The total chaos of the invisible micro one on fundamentally
43:02
creates the rock solid predictable order of our
43:05
practical engineering world. It directly connects the
43:08
invisible statistical behavior of billions of random atoms
43:11
directly to the tiny red needle physically moving on the
43:14
engineer's analog voltmeter. That is a genuinely beautiful
43:17
thought. To end on.
43:18
The terrifying ghost in the machine isn't some malicious
43:21
poltergeist after all. It is really just pure
43:23
statistics. Precisely.
43:25
It is just countless trillions of atoms blindly trying to find
43:30
their way back home to equilibrium.
43:32
Well, I think the engineer can finally power down his testel,
43:35
pack up his bench and go home to get some much needed sleep.
43:38
He earned it today. And hopefully you, the listener,
43:42
feel a little more equipped to tackle the thermodynamic ghosts
43:46
haunting your own machines and test rigs.
43:48
Just remember the golden rule, always physically check your
43:52
junction temperatures. Always check the junctions.
43:54
Isothermal blocks save lives. If you haven't already, give us
43:58
a like and follow. And please leave us a comment if
43:59
you enjoy the podcast. We love to hear from the folks
44:01
who make this possible. Until next time.