Latest / Mechanical Engineering Made Simple / (16) Heat Transfer Engineering - Thermal & Fluid Foundations .

Transcript

(16) Heat Transfer Engineering - Thermal & Fluid Foundations . · Mechanical Engineering Made Simple

• 0:00
  Welcome back to Mechanical Engineering Made Simple.
• 0:02
  Today we are undertaking quite a critical mission.
• 0:06
  We're doing a rapid, high impact deep dive, focusing on the
• 0:10
  absolute foundational concepts, the essential mathematical
• 0:13
  toolkit really, that undertines thermal and fluid sciences.
• 0:16
  We're aiming to extract those, you know, foundational Nuggets.
• 0:20
  You need to understand exactly how heat moves and how fluids
• 0:23
  flow. Yeah, this is absolutely
• 0:25
  fundamental territory. We're pulling together 4
• 0:27
  powerful areas from the source materials you've gathered.
• 0:30
  We'll cover the for thermal transport modes, the core
• 0:33
  conservation equations, mass and momentum, and then look at how
• 0:36
  engineers actually grapple with material properties.
• 0:39
  The Thermo. Physical properties.
• 0:41
  Exactly. And finally, the incredible
• 0:43
  power of dimensional analysis, which lets us scale things up or
• 0:46
  down. Right, taking insights from the
• 0:48
  lab bench to, say, a massive industrial process.
• 0:51
  Our goal here is pure leverage, isn't it?
• 0:54
  We want to quickly unpack these concepts, which can often seem
• 0:57
  dense. Very dense sometimes.
• 0:58
  Give you the context, the vocabulary, maybe those aha
• 1:01
  moments that make the real difference.
• 1:03
  Going beyond just knowing the equations to actually
• 1:06
  understanding the mechanics. Exactly.
• 1:08
  OK, let's unpack this first part.
• 1:09
  Then. When we look at how thermal
• 1:11
  energy actually moves around, the sources identify 4 basic
• 1:15
  modes, Conduction, convection, phase change, and radiation.
• 1:20
  Let's start with maybe the most intuitive one for solids or
• 1:23
  stationary fluids conduction. Right.
• 1:26
  Conduction, that's essentially heat diffusion is typically
• 1:29
  described by Fourier's law. But what's fascinating, and
• 1:32
  engineers do this almost immediately, is they translate
• 1:35
  this physical process into an electrical analogy.
• 1:38
  Using thermal resistance RCD. Precisely, RCD is defined as the
• 1:42
  length the heat travels divided by the thermal conductivity K
• 1:46
  times the cross-sectional area A.
• 1:48
  So L over KA. And it really mirrors Ohm's law
• 1:52
  for electrical current. It lets you visualize complex
• 1:54
  thermal paths, almost like simple electrical circuits.
• 1:57
  OK, and that analogy gets really interesting, doesn't it, when we
• 1:59
  move beyond just simple walls, Like consider one-dimensional
• 2:03
  conduction with internal heat generation.
• 2:05
  Think energy dissipation inside a chip or maybe a battery cell.
• 2:09
  Yeah, that's a great example. If that heat generation is
• 2:11
  distributed uniformly through the whole volume, like say
• 2:16
  evenly along the length of a current carrying wire, well, the
• 2:19
  sources reveal something critical for design here.
• 2:21
  Which is. The overall thermal resistance
• 2:23
  in that distributed case is only 1/4, just 25% of the resistance
• 2:28
  you'd get if the same total amount of heat was generated
• 2:31
  only right at the very center. Wow. 1/4 That's a huge
• 2:35
  difference. It is, and that four X Factor
• 2:37
  fundamentally changes how you approach thermal management.
• 2:40
  Because if you centralized the heat source, the peak
• 2:43
  temperature inside could spike, right?
• 2:45
  Potentially causing failure. Exactly.
• 2:47
  By distributing the heat generation, you keep those those
• 2:49
  potentially dangerous temperature peaks much much
• 2:52
  lower. Which explains why, you know,
• 2:54
  battery designers and chip engineers work so hard to spread
• 2:57
  that thermal load out. Precisely, and Speaking of
• 2:59
  design realities, we can't always assume perfect materials
• 3:03
  or perfect contact. When two real solids touch, even
• 3:07
  if they look smooth, we get this thing called interface contact
• 3:10
  resistance. ARCO.
• 3:12
  OK. So there's a measurable
• 3:13
  temperature drop right at that joint.
• 3:15
  Why does that happen if the surfaces look smooth to the eye?
• 3:19
  It gets down to the microscopic level.
• 3:21
  Real surfaces aren't perfectly flat.
• 3:23
  They have tiny peaks and valleys called asperities, roughness,
• 3:27
  and slope. Essentially, these disparities
• 3:30
  mean the actual solid to solid contact might only be say 1%,
• 3:35
  maybe even less of the apparent area you see.
• 3:37
  So the heat has to struggle to get across.
• 3:40
  It does. It has to transfer mainly by
• 3:41
  conduction through those tiny solid contact points, but also,
• 3:46
  crucially, through the fluid, usually air trapped in all the
• 3:49
  microscopic gaps between them. Right, so that's why we use
• 3:52
  thermal interface materials like thermal paste on a computer
• 3:55
  processor. Exactly.
• 3:56
  It's not just filling visible gaps, it's replacing that poorly
• 3:59
  conducting air in the microscopic gaps with something
• 4:02
  much more conductive. It helps bridge those
• 4:05
  disparities you got. It it drastically drops that
• 4:07
  Arcovalia that. Makes perfect sense.
• 4:08
  OK, shifting gears now what about moving fluids?
• 4:12
  That brings us to convective heat transfer, right?
• 4:15
  Convection is where the bulk motion of the fluid itself
• 4:18
  becomes a major player in transporting heat.
• 4:21
  We quantify this using the heat transfer coefficient H and that
• 4:25
  allows us to define a convective thermal resistance RCV very
• 4:29
  simply, just one over H * S so 1 divided by the product of the
• 4:34
  heat transfer coefficient and the surface area of the fluid
• 4:37
  touches. And can we use these resistance
• 4:40
  ideas for processes that change over time, like heating
• 4:43
  something up or cooling it down? Tell us about the lumped
• 4:45
  capacity model, ah. Yes, the lumped capacity model.
• 4:49
  It's a really useful simplification for analyzing
• 4:51
  transient heating or cooling, but it only really works well
• 4:54
  when the internal resistance of the object, its resistance to
• 4:58
  heat flow within itself, is negligible compared to the
• 5:00
  external resistance like the convection off its surface.
• 5:03
  So give us an example. Think of.
• 5:05
  It this way, Drop a very small, thin metal needle into cold
• 5:08
  water. It cools down almost instantly,
• 5:10
  right? The temperature inside is pretty
• 5:12
  much uniform throughout the process.
• 5:13
  Yeah, makes sense. Now drop a big thick steak into
• 5:17
  that same water. The outside cools quickly, but
• 5:20
  the center takes ages to cool down.
• 5:23
  The temperature is definitely not uniform.
• 5:25
  Right, big difference. The lumped.
• 5:26
  Capacity model is perfect for the needle.
• 5:29
  High internal conductivity, small dimensions, not so good
• 5:33
  for the steak. For those cases where it applies
• 5:37
  the system's time behavior, how fast it heats or cools is just
• 5:41
  defined by its thermal time constant.
• 5:42
  And that is? Simply the product of the total
• 5:45
  external resistance R like that convective resistance we
• 5:49
  mentioned and the objects thermal capacitance which is its
• 5:51
  mass times its specific heat MC. So R times MC OK.
• 5:57
  So all this talk about conduction, convection,
• 5:59
  resistance, it all fundamentally depends on the materials
• 6:02
  involved, doesn't it? Absolutely on their Thermo
• 6:04
  physical properties, as the sources call them.
• 6:07
  These are the raw data that fuel all our engineering
• 6:09
  calculations, and the sources seem to divide them into two
• 6:12
  critical camps. Yes.
• 6:14
  That's a very helpful distinction.
• 6:16
  First, you have the thermodynamic properties,
• 6:18
  sometimes called equilibrium properties.
• 6:20
  These include things like temperature, pressure, density,
• 6:24
  anthelfalpy and specific heat capacity ZP or CV.
• 6:28
  They describe the state of a system when it's stable at
• 6:31
  equilibrium. OK state properties.
• 6:33
  Right. The second camp is transport
• 6:35
  properties or non equilibrium properties.
• 6:38
  These describe the materials ability to move stuff, mass,
• 6:41
  momentum or energy when things are not in equilibrium.
• 6:44
  So this includes thermal conductivity Lambda viscosity
• 6:48
  ETA and thermal diffusivity alpha.
• 6:51
  And here's a point that jumped out at me from the sources.
• 6:53
  Something may be easy to overlook.
• 6:55
  The thermodynamic properties are generally considered much more
• 6:58
  accurate, much better defined by measurement, for common fluids
• 7:01
  at least, than the transport properties are.
• 7:03
  Is that right? That's absolutely critical, and
• 7:05
  you have to ask why. OK, why?
• 7:07
  Well, think about measuring pressure or density.
• 7:10
  It's relatively static, right? You measure the state, but
• 7:12
  measuring viscosity or thermal conductivity you actually have
• 7:16
  to induce a gradient. You need flow to measure
• 7:19
  viscosity or temperature difference to measure
• 7:21
  conductivity. You're observing A dynamic
• 7:24
  response. Exactly.
• 7:26
  And those dynamic measurements are just much more susceptible
• 7:29
  to experimental errors, to disturbances in the environment.
• 7:33
  It's inherently trickier. So the engineer constantly has
• 7:36
  to be asking, given this uncertainty in my input
• 7:39
  properties, how accurate does my final result really need to be?
• 7:43
  That's. The perpetual challenge How much
• 7:45
  accuracy is enough? Now, to deal with this,
• 7:47
  especially for fluids, modern analysis relies heavily on
• 7:50
  something called an equation of state, often based on Helmholtz
• 7:54
  energy. Yes.
• 7:55
  The Helmholtz energy formulation is incredibly powerful.
• 7:58
  It's typically composed of an ideal gas part and a residual
• 8:01
  part that accounts for real fluid behavior.
• 8:04
  So a 0 + O residual. And what's the big advantage,
• 8:07
  the leverage of using this approach?
• 8:09
  The leverage is mathematical consistency and efficiency.
• 8:13
  If you have this one master equation for Helmholtz energy as
• 8:18
  a function of temperature and density, you can then calculate
• 8:21
  all the other thermodynamic properties, pressure, entropy,
• 8:24
  specific heat, everything, just by taking its partial
• 8:27
  derivatives. Wow, OK.
• 8:29
  Instead of needing separate, potentially inconsistent curve
• 8:32
  fits for every single property, this one function generates them
• 8:35
  all. It's the mathematical bedrock.
• 8:37
  And to make this practical, engineers turn to tools like the
• 8:40
  NIST databases, right? They implement these high
• 8:43
  accuracy algorithm. Exactly.
• 8:45
  Databases like Reef Prop from NIST are indispensable, but what
• 8:49
  about when you're dealing with a new fluid, something not in the
• 8:52
  database? Good question.
• 8:53
  What then? That's where predictive methods
• 8:55
  come in, and a very important one mentioned is the method of
• 8:58
  Extended Corresponding states, or ECS.
• 9:01
  Extended corresponding states. How does that work?
• 9:04
  It's basically a clever scaling technique.
• 9:06
  It uses the well documented properties of a known reference
• 9:09
  fluid and scales them based on the critical temperature and
• 9:12
  critical density of your unknown fluid.
• 9:14
  So you can estimate the behavior of, say, a new refrigerant
• 9:18
  without having to do all the experiments from scratch.
• 9:20
  Precisely. It allows for quite accurate
• 9:22
  predictions, saving a lot of time and effort.
• 9:25
  It's widely used in chemical and process engineering.
• 9:27
  OK. Let's switch focus now to
• 9:29
  solids. How does heat move through them?
• 9:31
  In solids, thermal transport is mainly down to two things.
• 9:35
  In metals, it's largely the movement of free electrons,
• 9:38
  similar to electrical conduction.
• 9:40
  But in all solids, including insulators, it relies on
• 9:44
  molecular vibrations. Vibrations like atoms jiggling
• 9:47
  in the lattice exactly. And these vibrations are
• 9:49
  quantized, meaning they come in discrete energy packets.
• 9:52
  We call these packets phonons. Phonons so.
• 9:55
  You can think of a phonon as a sort of particle of vibrational
• 9:59
  energy moving through the crystal lattice.
• 10:00
  That's how heat actually propagates through an electrical
• 10:04
  insulator. And the differences in thermal
• 10:06
  conductivity that result from these mechanisms, they're
• 10:09
  staggering. Absolutely huge ranges the.
• 10:11
  Sources mentioned metals like silver way up at what, 427 watts
• 10:15
  per meter? Kelvin.
• 10:16
  Yeah, very conductive. While dielectrics insulators
• 10:20
  like paper can be down around .12 Willer MK A.
• 10:23
  Massive difference and. Then you have the famous Outlier
• 10:26
  Diamond up at 2310 Willow MK. Diamond is just incredible.
• 10:31
  It's because it's crystal structure is almost perfect.
• 10:34
  Very light atoms, very strong bonds, very stiff lattice so.
• 10:38
  The phonons can travel easily. Almost unimpeded, there's very
• 10:41
  little scattering. Now, unlike fluids where
• 10:43
  pressure is a big factor, a solids properties are mostly
• 10:47
  independent of pressure. But, and this is key, their
• 10:50
  thermal conductivity can be extremely sensitive to things
• 10:53
  like impurities or defects in the crystalline structure,
• 10:56
  especially at low temperatures. So.
• 10:58
  Even tiny flaws can disrupt those phonon pathways
• 11:01
  dramatically. Even parts per million
• 11:03
  impurities can sometimes slash the conductivity.
• 11:05
  It makes measuring and predicting solid properties
• 11:07
  quite challenging. How?
• 11:08
  Do we we measure them accurately?
• 11:10
  Then we use. Specialized techniques, things
• 11:12
  like differential scanning calorimetry or DSC are common
• 11:15
  for measuring specific heat and for thermal diffusivity.
• 11:19
  The flash method is a widely used modern technique.
• 11:22
  OK, moving towards the flow aspect now, the ultimate
• 11:25
  language here is mathematics, isn't it?
• 11:27
  It really is. We have these two core
• 11:30
  conservation laws that universally define fluid flow.
• 11:34
  The first one is the continuity equation.
• 11:36
  What does that represent? Fundamentally, it represents the
• 11:39
  conservation of mass. Keep track of the mass.
• 11:42
  Simple idea, but the equation can look complicated.
• 11:45
  In its most general vector form, yes, it accounts for unsteady 3
• 11:49
  dimensional flow where density can change.
• 11:52
  But the core idea is simple. The rate at which mass builds up
• 11:55
  inside a specific volume over time must be exactly balanced by
• 11:59
  the net rate at which mass flows across the boundary of that
• 12:02
  volume. Mass in minus mass out equals
• 12:04
  accumulation. OK.
• 12:06
  That makes sense. And if the fluid is
• 12:07
  incompressible, meaning its density is constant like water
• 12:10
  in many situations, then. It simplifies significantly the
• 12:13
  time dependent term dropped out, and the equation basically says
• 12:17
  that the volume flow rate in must equal the volume flow rate
• 12:20
  out of any given region. What goes in must come out.
• 12:23
  Got it. What's the second core law?
• 12:25
  That would be the momentum theorem.
• 12:27
  This comes straight from Newton's second law.
• 12:29
  Force equals the rate of change in momentum.
• 12:31
  FDMVDT when we apply this principle carefully to a fluid
• 12:37
  moving through a control volume, considering all the forces
• 12:40
  acting on it. Pressure forces, viscous forces
• 12:42
  from internal friction, body forces like gravity, and we
• 12:46
  relate those forces to the change in the fluids momentum as
• 12:48
  it passes through. We eventually arrive under
• 12:51
  certain assumptions at the Navier Stokes equations.
• 12:54
  The favours Navier Stokes equations.
• 12:56
  The governing equations for what conditions specifically?
• 13:00
  Typically for incompressible laminar flow with constant
• 13:03
  viscosity, they essentially balance all the forces acting on
• 13:07
  a fluid element against its inertia.
• 13:08
  It's change in momentum, OK. So if these equations are the
• 13:11
  definitive statement on how these fluids flow, why don't we
• 13:14
  just, you know, solve them for every engineering problem?
• 13:18
  If only it were that simple. What's?
• 13:19
  The catch the. Catch is, they are notoriously
• 13:22
  difficult to solve. They're non linear partial
• 13:25
  differential equations, meaning we often can't find exact
• 13:28
  analytical solutions, especially when the flow becomes turbulent,
• 13:31
  which very often does in engineering applications.
• 13:33
  So. Even with powerful computers
• 13:35
  running CFD computational fluid dynamics, it's still hard.
• 13:40
  It's. Still computationally very
• 13:42
  expensive and requires careful setup and interpretation.
• 13:46
  The difficulty in solving Navier Stokes directly is precisely why
• 13:50
  engineers rely so heavily on another powerful tool, which is
• 13:54
  dimensional analysis, right? When you're faced with a complex
• 13:58
  problem, maybe flow in a pipe, heat transfer from a fin
• 14:01
  involving say 8 or 10 different variables, Velocity, diameter,
• 14:05
  density, viscosity, pressure drop, roughness, specific heat,
• 14:09
  it gets overwhelming fast. Exactly.
• 14:11
  Trying to experimentally determine how pressure drop
• 14:13
  changes when you vary each of those things independently would
• 14:16
  be a nightmare. So dimensional analysis helps
• 14:19
  cut through that complexity using the Buckingham theorem,
• 14:22
  yes. Buckingham is honestly an
• 14:25
  incredibly elegant piece of mathematical leverage.
• 14:28
  It provides A systematic way to group all those initial,
• 14:31
  potentially numerous physical variables into a much smaller,
• 14:35
  more manageable set of dimensionless groups.
• 14:38
  We call them π terms How? Does it know how many groups?
• 14:41
  The theorem tells you the number of independent dimensionless
• 14:44
  groups you need is equal to the number of original variables
• 14:47
  minus the number of fundamental dimensions involved, like mass,
• 14:51
  length, time, maybe temperature. Let's use the classic example
• 14:54
  from the sources predicting friction loss or pressure drop
• 14:58
  in fluid flow through a pipe. Perfect example, you might start
• 15:01
  with say 6 variables, the pressure loss per unit length,
• 15:04
  the fluid velocity, the pipe diameter, the pipe roughness,
• 15:07
  the fluid density and the fluid viscosity. 6 variables, and.
• 15:10
  The fundamental dimensions involved are mass, length and
• 15:13
  time, SO3 dimensions. 6 variables -3 dimensions means.
• 15:17
  You only need 633 eagles, 3 independent dimensionless groups
• 15:20
  to describe the entire phenomena.
• 15:22
  And those groups are? Typically they turn out to be
• 15:24
  the friction factor F which represents the dimensionless
• 15:27
  pressure drop, the relative roughness Ed comparing the
• 15:32
  roughness height to the diameter, and the reciprocal of
• 15:35
  the Reynolds number which involves velocity, diameter,
• 15:38
  density and viscosity, so. Instead of needing complex multi
• 15:41
  dimensional plots relating pressure drop to velocity and
• 15:44
  any diameter and density and any viscosity and need a roughness
• 15:48
  you. Just need one chart like the
• 15:49
  Moody chart which plots the friction factor F against the
• 15:53
  Reynolds number with different curves for different relative
• 15:56
  roughness values. That's the power One set of
• 15:59
  experiments on one pipe with one fluid can suddenly be used to
• 16:03
  predict friction loss in a completely different pipe with a
• 16:06
  different fluid at a different velocity, as long as the
• 16:09
  dimensionless groups, Reynolds number and relative roughness
• 16:12
  are in the right range. Exactly, it allows for scaling
• 16:14
  and generalization of experimental data.
• 16:17
  And this is fundamentally why dimensionless numbers like the
• 16:20
  Nusselt number for convection, the Prantle number relating
• 16:24
  momentum and thermal diffusivity, and the Reynolds
• 16:26
  number for flow regime are the absolute lingua franca of
• 16:30
  transport phenomena in engineering.
• 16:33
  They logically group the interacting variables, simplify
• 16:36
  complex behavior, and provide the essential framework for
• 16:39
  correlating experimental data and applying it broadly.
• 16:43
  They connect the fundamental physics to practical engineering
• 16:46
  charts and correlations. Hashtag tag.
• 16:48
  Outro. So we've covered quite a bit of
• 16:50
  ground today. We navigated the four modes of
• 16:52
  heat transfer, conduction, convection, phase change,
• 16:55
  radiation, dug into the critical thermophysical properties of
• 16:58
  fluids and solids. Including that fascinating world
• 17:01
  of phonons and the power of the Helmholtz energy approach,
• 17:04
  right? And we touched on the core
• 17:05
  conservation equations governing flow and the immense practical
• 17:09
  leverage we get from dimensional analysis.
• 17:11
  It really. Highlights how interconnected
• 17:13
  these concepts are. But it also raises, I think, an
• 17:15
  important question for you, the practicing engineer or the
• 17:18
  student to consider. Oh, what's that?
• 17:20
  Well, we rely so heavily on these elegant mathematical
• 17:23
  models like Navier Stokes and these powerful predictive
• 17:26
  concepts like extended corresponding states.
• 17:29
  But as we discussed, these models are entirely dependent on
• 17:32
  the accuracy of the underlying Thermo physical property data we
• 17:36
  feed into them viscosity, thermal conductivity, density,
• 17:40
  specific heat and so on. And we know measuring those,
• 17:43
  especially transport properties, can be tricky and materials,
• 17:47
  especially solids, vary due to impurities.
• 17:50
  Exactly. So the question is, considering
• 17:52
  these inherent uncertainties in the data, how much accuracy is
• 17:56
  really enough for practical engineering design?
• 17:59
  When does the quest for more precise property data yield
• 18:02
  diminishing returns compared to, say, refining the model itself
• 18:06
  or simply using appropriate safety factors?
• 18:09
  That's a really interesting point.
• 18:10
  How does that continuous refinement of property data
• 18:13
  fundamentally change our designs versus refining the models or
• 18:16
  just accepting uncertainty? It's a constant balancing act in
• 18:19
  engineering. Definitely something to chew on.
• 18:21
  Indeed, lots to think about there.
• 18:23
  Until the next equation, I'll see you in the flow.