Hacker News new | past | comments | ask | show | jobs | submit login
Airfoil (ciechanow.ski)
2544 points by todsacerdoti 8 months ago | hide | past | favorite | 296 comments



It's pretty interesting that many airfoils used in aircraft design were derived by NACA in the 1920s and 1930s [1]. You'd think that with modern computer software it would be possible to design better airfoils, but apparently, those shapes have already been mathematically perfected by hand and by experiment. So nowadays if you want to design a plane you can just look up the desired NACA airfoil from a table based on the speed, air pressure, etc that you require.

[1] https://en.wikipedia.org/wiki/NACA_airfoil


Not quite true! Modern airplanes are way more complex. First of all, all modern airplanes have supercritical airfoils which go back to the 60s and 70s. Secondly, the airfoil of the wing root is typically different than the wing tip. Finally, new composite wings are adaptive during flight. They change their shape slightly to maximize efficiency.


Case in point would be modern gliders (sailplanes). One simple parameter that describes their aerodynamic performance is the maximum achievable Lift/Drag ratio, and that dimension-less ratio has climbed from ~30 in the 1960s to as high as 75 today. That means modern gliders can, using the same altitude/energy, go over 2 times further horizontally. The L/D is not the ultimate decider of performance but it is quite representative of the aerodynamic performance improvements.

BTW, all lift based flying objects have an L/D ratio (which depends mainly on the airspeed), this includes birds, fighter jets, commercial airliners; and the discrepancies can be pretty interesting. For example if one looks at the L/D of the Concorde vs a subsonic jet it becomes clear why it was so damn expensive to operate. Or why the U-2 looks like a glider :). I cannot find any aerodynamic performance data on any famous long endurance (>24h) unmanned drone, but I bet it's rather high as well.


> Concorde

Another good example is the space shuttle. It does actually glide back down. But it glides like a brick at first (1:1 during its initial braking into the atmosphere), and then like a less dense brick (2:1 while it's still supersonic), and then like a brick with shitty wings (a whopping 4:1 or whatever on final approach). Which is about what the Concorde is during landing, 4:1, yea.

Pretty crazy stuff

(Obviously the space shuttle was a tradeoff for, you know, getting it into orbit via rocket)


Your numbers are right but your analogies are misleading. I get “glides like a brick” is hyperbole, but you’ve added enough detail I can see people taking it seriously.

A brick’s L/D is much worse than 1:1. I’m seeing people say 1:10 online, but I can’t find a source and I think that’s incredibly high. A real brick is going to tumble and essentially not make any lift.

A less dense brick will have the same L/D. L/D is about the shape, not the mass.


Space Shuttle pilots themselves referred to it as “The Flying Brick,” I think that is mostly what they were referencing. It was a term of endearment :)


"The F4 Phantom...America's proof to the world that with a big enough engine, even a brick can fly"

https://www.youtube.com/watch?v=XqUgUgiToNs


Yea sorry, this was it. I didn't mean it literally has the glide ratio of a brick

I mean it looks like a brick and it flies


Despite knowing about its nickname… I found myself picturing a paraphrasing of the classic physicist’s “spherical cow in a vacuum”…

“Steerable brick in an atmosphere”… or the slightly more accurate “orientable brick in an atmospheric reentry regime”…


If you want a real "orientable brick in atmospheric reentry regime", check out how they steered the Apollo capsules back down. Kinda bonkers, but worked - they made it asymmetric and then rotated it depending on if they wanted to go down faster, slower, left, or right. Obviously it can't fly, and its glide ratio is actually brick-like unlike my facetious description of the shuttle. But it worked enough.

Kinda like a single control plane missile that spins (rolling airframe?), except... without the control plane lol.


I’m trying to get my head around the L/D being independent of mass. Does lift scale with airspeed at the same rate as drag? Or is L/D only considering lift-induced drag (whatever the term is) and not total drag including parasitic drag?


Roughly speaking, both lift and drag are proportional to v^2 for a given geometry.

Neither lift nor drag has anything to do with mass. They are entirely determined by the surface of the object, and are not affected at all by the interior properties, including density.


The way I've had it described is that when two objects of the same shape pitch for optimal glide (i.e. highest L/D) then the heavier one will reach the ground sooner (go faster), but both will take the exact same path and land the same distance away. In other words, same L/D.

This is not the explanation you are looking for, but "aha the heavier object takes the same path but drops faster" was what made me okay with L/D not depending on weight.


L/D changes with angle of attack. You can have a different airspeed at the same angle of attack and the ratio does stay the same. I think the Wikipedia page gives good descriptions.

Something a bit misleading done generally is aircraft don’t have one L/D, they have many, depending on angle of attack. When you see one number, it’s usually the best one.


Aren't they comparing high-supersonic to supersonic to subsonic?

I thought the point was that aerodynamics change from one domain to the next as shockwaves cause flow separation or eddies on or behind surfaces.


>> tumble and essentially not make any lift

Tumbling itself can produce lift. The difference in drag between one side and the other can result in net pressure differences for a moving object. This is the basis of many baseball pitches. Spin a brick fast enough and it might just be able to climb if thrown horizontally.

If static airfoils are complicated, try looking into airfoils that rotate or otherwise move in relation to airflows. A Russian engineer once said that all problems in aerospace are placed on the tip of every helicopter blade.


The Magnus effect is super confusing to think about. Basically, to provide lift, the brick or ball or whatever would need to be "rolling" backards, like a wheel. In baseball, e.g. fastballs are usually thrown in such a way as to "roll" backwards, which causes them to climb. Curveballs are accomplished via topspin. Sinkers roll forward. You can even tune this behavior by making the surface of inconsistently "sticky" to the air, so flow of air is more or less affected by the objects rotation. This is why licking the baseball is against the rules.

Curiously, the rotation can also lend the ball's path greater stability against changing air currents/densities and crosswinds. Knuckleballs are famously hard to throw because they have very little spin, but they are also notoriously hard to hit because the trajectory is so subject to the vagaries of airflow between pitcher and batter.

This is to say nothing about when the axis of rotation is predominantly parallel to the direction of travel (e.g. rifle bullets and American footballs), where the Magnus effect effects the rotating objects ability to continue to rotate parallel to the direction of travel. Get it right and the spin makes the path more stable, but get it wrong, it becomes less stable. The hows and whys of that are beyond my understanding of fluid dynamics, but its fun to think about how complicated it can get.

edit: got my spin directions confused


I guess you get maximum L/D out of a brick by giving it some serious backspin (which is a stable configuration so it might be able to maintain it on the way down) to set up the lifting circulation around it. But would this count?


I saw the space shuttle land once. From my perspective, it seemed to drop like a rock (fast!) and then as it got closer to the ground, it started "flying". I'd never seen anything like it.


The flare-out is really difficult to get right, from what I’ve read. As soon as you start leveling off, air speed is going to drop really fast, and you have very little time to get that bugger on the ground before you get in a stall. They used to practice using a modified jet, flying from high altitude to a landing with thrust reversers engaged all the way!


For whatever reason the way I think about this phenomenon is like autorotation in a rotary craft.


I personally give the space shuttle a little slack on the aerodynamics department when I remember that it enters the atmosphere at nearly Mach 25, and is big enough to produce two separate sonic booms. It was probably a very wasteful design (burdened by the military requirements), but it's like nothing else we've ever built!


The case with gliders and U2 and the max L/D is due to the wing aspect ratio (look up the formula for drag polar). Modern aircraft have much higher L/D because they have long skinny wings and these wings are possible because we moved from aluminum to carbon reinforced composites.


NACA airfoils aren’t so much a numbered set of standard, tested designs as a useful set of mathematical curve formulae for making airfoil-like shapes, and describing them using parameters.

NACA published empirically determined wind tunnel performance numbers for selected parameters, which was useful research but not a declaration of ‘these are the good values, you should only use these’.

It’s a bit like saying all satellites follow TLE orbits derived by NASA/NORAD in the 1950s - they do, but only because that’s just a standard way of writing down the orbital elements that describe a particular ellipse, not a catalog of ‘known good’ orbits.


my recollection was that the P51 wing used a NACA airfoil and it had low drag properties not commonplace at the time


"better" airfoils are used in experimental craft design. for example mark drela wrote and used xfoil to design wings for mit's project daedalus, a human powered long distance flight aircraft. this is the case where, like sibling commenter stated, you need that extra % to get better performance characteristics. you can still run xfoil, it's a delightfully oldschol fortran program.


I have a friend whose PhD is in computational flow dynamics, as applied to airfoil design. He works almost exclusively in fortran (which is wild to me, for someone under 35, but I guess its the "industry" standard). I just asked him about xfoil and he observed that there are more modern programs for (as he put it) more "realistic/complex" designs, but said it was a good starting place.


Because Fortran was the industry standard for any heavy scientific calculation (like aerodynamics or nuclear bombs), the Fortran compiler has been optimised to death... And thus Fortran is still the industry standard.


oh i'm sure state of the art has advanced since then. i have the necessary physics background, but it's not otherwise my domain: i once used xfoil to design an airfoil for an autonomous model glider as a hackerspace project back when i had free time for things like that many years ago. the glider was also loaded into x-plane to develop and test the autonomous part. so whenever various experimental aircraft projects popup, i'm likely to look into them, and then also notice the peculiar foils they use.


It really depends. Up until the 90s Fortran was completely dominant. These days a lot of people have moved to C++. Important open source codes such as OpenFOAM and SU2 are in c++.


FORTRAN is easier to optimize than something like C so is much used for numerical computation


NACA and the other published airfoils[1] are generally a good starting point for hobbyist/RC folks. However if you want to eke out that last 5% bit of performance (ie. you are a company/institution), you would start with one of the above airfoils and optimize them to fit your flight envelope & mission profile. Here's a neat video of optimizing a round profile into an airfoil optimized for supersonic speeds [2].

[1] http://airfoiltools.com [2] https://www.youtube.com/watch?v=FHYTBguMfWc


For gliders the naca airfoils have been abandoned around 1970, when the first glasfiber composite gliders were made. We mostly use airfoils from German professor Wortmann (FX) , Quabeck (HQ) or Boermans (DU). The naca airfoils are still used in wind turbines though.


Eh, it's more like you can get to 90% of where you want to be with the 100 year old airfoils (though several of the other series are quite a bit newer).

https://aviation.stackexchange.com/questions/20798/are-naca-...

>You'd think that with modern computer software it would be possible to design better airfoils, but apparently, those shapes have already been mathematically perfected by hand and by experiment.

No, modern computer software indeed does better, but there's not a whole lot of room to do better, small changes to bump performance a percentage point or two. These are optimizations which can be (and are sometimes) skipped for many commercial projects.


The big variation now though is that the airfoil shape varies quite a bit from one end of the wing to the other.


Different airfoil shapes for the root and tip were common already in WWII era planes.


Not really, they're a nice first step though, or if you require something "good enough".


This is one of the best thought out UX I've ever seen. It's extremely well laid out and simple to navigate through, all the design choices are very meaningful, UI elements (like the unit conversion) are available inline when you need them...

Not to mention the content itself is great.

I'm taking notes.


You probably want to check out the rest of his site, which is pretty much on par with this: https://news.ycombinator.com/from?site=ciechanow.ski


His mechanical watch piece is really something to behold.


I'd love to know how long it takes him to make each one of these "interactive essays"


I'd love to read about how he does it. How talented do you have to be to not only learn all this stuff but also model them in 3D and simulate them in real time and interactively?


Guessing from personal experience, I would not be surprised if 70 % of the domain knowledge is learned in the process of writing and modeling.

Whenever I encounter a tricky subject I'm having a hard time learning, I start writing an article explaining it to someone else. It really forces me to confront the gaps of my knowledge because I can see so clearly that "Wait, I can't explain what happens between these two steps here. What am I missing?"


I would also be interested in knowing which software is used?


Hah. That question comes up on every single one of his explainers. The answer is none; he hand-crafts the JavaScript and WebGL shaders. From the design language, I’d guess he has built up a library of templates and snippets to draw from by now, but all in all, each of these pages is a bespoke work of art.



Looks like pure JavaScript using the WebGL Canvas.


A long time would be my guess but writing interactive articles is much more engaging and addicting. We've started adding interactive stuff to our docs and it is really engaging so the spent hours just fly by!


Also check out https://pudding.cool if you’re unfamiliar and enjoy extremely high effort visualizations alongside editorial and educational text content.


I grew up duck hunting and learned intimately how ducks use their wings and the variations of shapes at different velocities as they slow down to land on the water. I also grew up boating and swimming and have a likewise similar understanding of paddling, tracking a canoe straight, and using boat motor trim to "get on the plane".

I guess I struggle with articles like this because it's already so intuitive as a mix of air and fluid dynamics. In fact, fixed airfoils are so boring when you see what a duck can do.

https://www.youtube.com/watch?v=-3CVZYY8xS4

So for all the fancy physics talk, this duck is literally just paddling air with his wings. The same physiology I use to stay afloat when treading water while swimming.


But when that fancy duck wants to get to Paris in a hurry, it still has to hop on a fixed-wing Concorde like everyone else.


Give it a strong selective pressure towards speed and a few million years and you will have your supersonic duck.


Intercontinental ballistic mallards? Just another trillion-dollar boondoggle from the military-ornithological complex.


Reverse dragooning by expelling fiery farts made of hypergolics fueled by fantastic fermentation?


Something like that is featured in Terry Pratchett's Discworld series of books. (Don't spoil the book it's in!)

GNU Terry Pratchett


There's also that Harry Potter spinoff movie in which he becomes a supersonic corpse with a farting jet-butt in order to visit some pretty redwood parks: https://www.inverse.com/article/17614-how-swiss-army-man-mad...


Something similar already happened in reality in another evolutionary chain:

https://en.wikipedia.org/wiki/Bombardier_beetle

(on a much smaller scale)


Hummingbirds go to mexico to vacation a lot more than I do. Imagine if humans had a manadatory 2000 mile trip and 6 month layover twice a year to survive.

Once you have technology that enables flapping type motion, it's opening up the applicable physics to like 6 degrees of freedom versus zero in current wing technology (fixed = 0 degree of freedom); much more complex and interesting to study.

How else will we move toward ornithopter style wings, or vehicles that can hover via wing movement.


Rotating wing aircraft have no problem hovering mid-air. All we need are a handful of breakthroughs in battery tech.


Rotating wing aircraft have problems with supersonic flight - and the wing (rotor) itself reaches supersonic much quicker than the aircraft itself, thats why helis are usually slow, compared to aircraft.

I guess “supersonic wing flapping” would have similar problems, but maybe there are more clever solutions than can be modeled, with so much degree of freedom?

And is “supersonic flapping” even possible?


The arctic tern flies from the north pole to the south pole.


I once ate duck in Paris.


That is a remarkable story...


> this duck is literally just paddling air with his wings

Grossly speaking, sure. But I feel like this simplifies away a lot of the interesting bits. It's not as simple as, say, someone on a canoe paddling. Why is the duck's wing shaped just so, and not another way? Why does it move its wings just so instead of another way?

I'm reminded of an analysis of fruit fly wings, showing how they re-capture energy from the air when flapping[1]. Maybe the duck is doing similar; I don't know.

Of course, these animals make it look easy, thanks to millions of years of evolution (:

https://www.nature.com/articles/s41598-021-86359-z


Fixed airfoil physics become really important at very high speeds.


Or if your Cessna isn’t equipped with a flapping system.


Amusingly, the flappy Cessnas were designed with the engine mounted right in front of the windshield.

Where else could it possibly go?

https://gallery.vtol.org/image/P2xYJ


I think you might be experiencing a bit of a dunning-kruger effect

Also, in my experience there’s a huge difference between having an intuition for something and having an understanding of something to the point where you could model it.


For an example of a flat-ish airfoil that performs well enough for model airplanes (and is easier to build than a NACA & co airfoil), see the KFm airfoil family:

https://en.wikipedia.org/wiki/Kline%E2%80%93Fogleman_airfoil

Very useful when making model airplanes out of foamboard.


Huh, odd. I was under the impression that for swept/delta winged paper airplanes one wants a smooth top surface to encourage attachment and any steps on the bottom to provide decalage. (I.e. the area ahead of the step acts as a canard-like surface.)

Is this an airfoil that works for tailed aircraft but not tailless ones, perhaps?

Edit: I just skimmed the book on paper planes by KF and indeed they are using the variation with the step on the bottom for their paper planes.

I'm actually even more surprised now. How on Earth did they manage to patent the idea of reflex on a delta wing to give a tailless plane stability? This seems like the thing that (a) was known since early human-carrying gliders, and (b) implicitly discovered by anyone that folds a lot of paper airplanes. I will definitely read their book in more detail.


Attachment isn't as much of an issue with paper planes since the small size and low speed give much more favorable Reynolds numbers.


Could you elaborate on that, please? This is just at the edge of my understanding and I'd like to learn more.


You can see the effect in the second animation of the linked page, but the basic idea is the lower the Reynolds number, the less likely for flows to separate and become turbulent. Shrinking the scale and slowing the airspeed both lower the Reynolds number, so paper planes have vastly different aerodynamics to full size aircraft.


And even further, insects have also vastly different aerodynamics, which explains why flapping flat wings works at insect-size and not at bird-size (nor plane-size).


Oh no. A while ago I stumbled on this guy's page about mechanical clocks - a subject I was never interested in - and was forced to spend the rest of the day studying clocks.



I wish every presentation on how planes fly started with an actual flat plane. A wing that has a flat crossection. I think the shape of the airfoil of the wing is absolutely distracting and prevents people from understanding what is really happening.

Every person who ever stuck a flat object outside the window of a moving car knows that you do not need a fancy shape to have lift.

And so many people are stuck thinking that the shape of the airfoil is responsible for the plane to be able to fly, supposedly because the air needs to run a longer way around the foil above the wing than below the wing. And this somehow causes pressure difference due to Bernoulli law and this is what keeps the plane up. Which is almost total BS because planes can obviously fly inverted.

Now I admit I only skimmed the article, and although the animations are beautiful, I am missing what really is key to understanding of what is happening.

I am looking for a bigger, far away view of the wing and showing what happens to the air BEHIND the wing.

Because how the plane really works is as it flies forward, it diverts large masses of air downwards. It pushes off of air.

Part of the air is diverted by the lower portion of the wing, but the much larger portion of lift is generated by larger masses of air above and behind the wing. Those can be thought as being sucked down behind the wing (if you look at it from the point of view of a stationary air mass, not from the point of view of the wing).

And the main role of the airfoil is to keep that mass of air behind the wing stuck to the airfoil at wide range of angles and speeds as possible, because a flat sheet is very poor at doing this.


Yea, I agree and try to explain it this way to friends. Airfoils help, but it's ultimately just the wing pushing air down and why planes can fly upside down.

FWIW, aerospace engineering degree, used xFoil, did tons of fluid sims, etc.


And it's an "even more wrong" explanation than the "lies for children" diagram used in school physics class.

For reference, actual "proper" discussion of lift in textbooks on aerodynamics have tendency to start with a sphere/cylinder.


Do you have any recommended reading on this topic? I'd like to brush up.


Understanding Aerodynamics: Arguing from the Real Physics by Doug McLean, a former aerodynamics Technical Fellow at Boeing Commercial Airplanes.


Bill Beaty's site was the one that opened my eyes to these misunderstandings: http://www.amasci.com/wing/airfoil.html

If the diagram shows lift but doesn't show the air being directed downward after leaving the tailing edge of the wing, I basically stop reading. That's the whole thing.


Thank. You. That's exactly what is missing and that's exactly what I have mentioned in my... highly criticised comment. It just shows how pervasive the misconception is.

If you take a step back there is a simple way to think about this. In order for the object to stay up there, there needs to be equal and opposite force from some other body. What is that other body? It is the mass of air that is being directed in the opposite direction of the lift force acting on the plane.


I think the mistake this site is making is trying to model out pressure changes that would create an airfoil-like shape of airflow. That's not how wings work. The pressure at the wing's leading surface is infinite because there's a metal skin there. You don't need a low pressure zone sucking air up 1m above it to explain that.

This view also implies that most of the lift is happening on the very front edge of the wing which I doubt is accurate otherwise we would have very skinny wings.

It's super frustrating when wrong is very pretty.


There used to be a good one from NASA, written for K-12 but 100% adhering to actual science not "lies for children".

EDIT: This is a good starting point for the frankly awesome material from NASA Glenn Research Centre: https://www.grc.nasa.gov/WWW/K-12/VirtualAero/BottleRocket/a...

Unfortunately it partly bitrotted due to using java applets for interactive demos, but I think most of it is still reachable - I'll try to find it later when I'm at the desk.

Personally I learnt from a 1980 book that was still part of mandatory reading for glider pilot course in Poland in 2005.


There is no lift on a sphere or cylinder without rotation dude. The whole point of parent post is that the "proper" discussion does not inlay a good intuitive understanding of lift, which in my opinion, should start with "push air down to go up".


Yes, there's no lift.

But there's quite different flow and drag around it, which was used as opening for for adding rotation (which would add viscosity effects including lift from rotation) and other changed shapes in better way than starting with flat plane.


Why is it wrong?


To me the most intuitive and practical mental image is imagining two large bubbles of lower pressure above the wing that hold the wing up by suction (you can see those literally as condensation under certain conditions). As you increase the angle of attack the bubbles get larger and stronger, until the angle is so large that they “break off” and the wing stalls.


I once found an explanation that finally made it clear to me why the shape of the airfoil can create lift. Yes, the air above the wing needs to travel a longer distance with the typical section used in wings, which means that it goes faster than the air below the wing. It also leaves the wing moving downwards - and when this downward-moving, faster flux of air meets the slower one from below, the result is that a mass of air is pushed downwards - exactly as needed to lift the plane, as you correctly said.

As the article says, you can have lift by just changing the inclination of a symmetrical airfoil, but an asymmetrical one can generate lift even without inclination (and with lower drag). The article also explains that acrobatic airplanes have symmetrical wing sections exactly because they need to be able to fly just as easily inverted.


> Yes, the air above the wing needs to travel a longer distance with the typical section used in wings, which means that it goes faster than the air below the wing.

Both of these sub-clauses are true, but the "which means" connecting them aren't. There's no law of physics saying a fluid that has a longer path ahead of it speeds up in anticipation.


Isn't there an even more basic explanation: If incoming air hits a flat surface at an angle, and is deflected downwards, then by the law of action and reaction, the surface itself moves upward.

As a child, I quickly outgrew the airfoil explanation when I realized this.


That's exactly what is happening. But it is also not enough for the airplane to fly.

In a normally flying airplane, the wing compresses and pushes an amount of air under its wing. But there is actually even greater amount of air sucked down by the region of underpressure created above the wing and by the laminar flow directing it downward. Here, the drawing at the top of the page makes it clear: http://www.amasci.com/wing/airfoil.html

When you have a stall condition, what happens is that the air below the wing is still being compressed and directed downwards, but the air above the wing becomes turbulent and "unsticks" from the surface of the wing. Rather than being nicely directed downward, it just dissipates a lot of energy in turbulent motion that is not directed in any particular direction.

This turbulent air not only ceases to provide lift, it also prevents the air from below the wing to be directed downwards efficiently.

The main job of an airfoil isn't to create a pressure difference, it is to create conditions for the air to be laminar at as wide range of speeds and angles of attack as possible to make the plane nicely behaving and possible to takeoff and land. It is super critical for landing as you need to have higher the angle of attack the slower you fly and all planes essentially are driven as close to stall as possible during landing. Similar happens at high altitudes and high speeds, but for a bit different reason (read up on "coffin corner" if you are interested in that sort of thing).


Great explanation. In addition, flaps make the wing able to provide lift at slower airspeed at the cost of efficiency. Perfect for takeoff and landing.


Yes and no. The thing you describe happens, but it's not enough to explain the amount of lift generated by a wing, because a surprising amount of air hits also the top of the wing! The difference in pressure between top and bottom wing surface is just a few percent.

The reason wings produce significant lift anyway is that they deflect air far beyond their surface. Air several metres away from the wing is also deflected downward, even though it doesn't actually hit the wing itself.

So yes, Newton's third law is involved, but in a "spooky action at a distance" form, where the wing somehow manages to deflect a bunch of air it doesn't even touch!


My dad who worked on wind tunnels just flat said you can either integrate the pressure over the surface of the wing or the momentum change as the air passes over to derive the amount of lift.

Both give exactly the same results and are convertible mathematically.

For wind tunnel work it was easier to measure pressures.

I'm with you I don't think the standard hand wavy explanation gives you the ability to attack the problem mathematically. So it's basically wrong.


My understanding is that "which means" only makes sense with the assumption that what is being studied is the laminar flow of an incompressible fluid (which was described as a fair assumption for air and a wing at subsonic speed). But thinking more about it, it's probably right that this isn't about the fact that the air above needs to travel a longer distance, which would also be true for a concave wing section, but the fact that the layers immediately above the wing need to travel the same X distance through a thinner Y section - as in a tube which becomes thinner. Which forces the fluid to go at a higher speed, and have a lower pressure.


Unfortunately, your explanation is entirely wrong... and you're attacking a "lies to children" simplification with your mention of "needs to run longer way around" bit.


Well in defense of the GP, the "planes can observably fly upside down" point (and its close cousin the "flat wing cross sections can fly too" point) is a good one, this pokes holes in the usual two-dimensional "the air goes faster on top" themed explanation that omits any discussion of vortex shedding/third-dimensional effects.


Oh, to be quite honest, I loved trolling my high school teachers with "your explanation fails, here is a real world airfoil, please explain it" and I would draw a symmetrical airfoil or - for extra trolling - a trapezoid one. (At that point I had already flown solo)

But the same I found myself unable to pass by someone pushing "flat plane at an angle".


Did you have to write the lie in an exam to pass too?


Fortunately the exam questions that involved lift in high school were simple enough they didn't trigger "you're wrong and the textbook is wrong" response XD

Fortunately my exams were open ended not "fill in the circle in answer sheet" so worst case I'd have written a more complete answer and fought it out.

Worrying about having to fight against "answer key" is part of why only one person (and only on a lark) took computer science on Matura exam in my class - which was CS-math-physics focused one


You mean it is more akin to those grills you position to control your AC pushing the air in a certain direction. But with just one surface you get the AoA too high problem. Hell I am gonna stick my hand out next time in a car (being safe about it!) and see the stall angle of my hand.


> planes can obviously fly inverted

Many (most?) planes cannot sustain inverted flight.


IIRC that is due to other issues, not aerodynamics.

For instance, the engine no longer receiving oil at negative 1 g, or fuel, as the system is designed for gravity flow.

Stunt planes and airplanes capable of long inverted flight need special oiling and fuel systems to keep the engine from starving from either.


Fascinating! Commercial airliners have also optimized their engines enough they don't have the power budget to make up for the difference in lift.


I agree with the other commenter that the specific shape of the cross section of the wing is overemphasised in almost all material including this one. Any shape longer than it's thick will, at a reasonable angle of attack, provide lift.

This article did provide a barn door model also, but it was quite far down.

The shape is mainly about efficiency and increasing the range of reasonable angles of attack, and then further nuances.


It's amazing how the article did such an incredible job building a deep understanding of how the airfoil works, yet you managed to completely miss that and find something to so small to critique.


To be clear, the article was amazing. That has already been said multiple times by others so if I left a comment saing just that I would contribute nothing. Besides, the size of the criticism (in this case small, as you point out) is an even better measure of quality than number of fawning comments.

I also publish articles (though nowhere near as good or ambitious as this one) online and the comments I look forward to most are the constructively critical ones. They are the reason I publish in the first place.

My only goal of giving and receiving constructive criticism is to improve our collective understanding of the world. There's nothing sinister or ill-natured about it as another commenter suggested.

(This extends to comments as well. I really appreciate you prompting me to check my tone.)


This made me think of how I hate Youtube comments. All the high-fiving positive ones end up on the top and not the ones that provide an opportunity to learn or think critically.


bro just stop


I think many of these kinds of comments are driven by a form of insecurity. They subconsciously wish they had written the article and are envious of the attention the author is receiving… so they find whatever small nitpick they can in order to tear it down.


Sorry for my low-value comment, but I think it is appropriate here. Doing a psychoanalysis of OP does not really add to the discussion meaningfully. Same applies to parent comment.

https://news.ycombinator.com/newsguidelines.html

> When disagreeing, please reply to the argument instead of calling names


I thought they made it very clear and talked at length that the shape isn't the key important factor. They do also then go on to talk about the benefits of different shapes and why they are chosen.


You gotta wonder why the Wrights spent so much time optimizing an airfoil. You could have been there to tell them to use a barn door wing and flat plate prop to save them a lot of time.

I have noticed this "it works or it doesn't work. Everything else is nuances." binary thinking among SWEs. It's odd.


Don't get me wrong -- for practical flight it is really important to expand the reasonable range of angles of attack because angle of attack is one of very few ways one has of controlling the aircraft.

But for explaining how lift appears, it is an irrelevant detail.

The purpose of modeling is not to mimick reality at high fidelity but to focus attention on the o parameters that matter for a specific situation. When you change the situation (going from explaining how lift happens to trying to fly) it is not surprising to have to switch to a different model.


An odd thing to say for a SWE. There's huge differences in the quality of apps, both from a user perspective and an internal/dev perspective.


effect is amplified on hn


This article shows how a flat plane can create lift. It doesn't start with that, it starts with a familiar airfoil, deconstructs it until it reaches the flat plane making lift, and then builds back to showing why airfoil shapes are used when simpler flat planes still make lift.


Regardless of all the exacting comments, this is one of the most astonishing accounts and content on a topic I've seen for years; beautifully delivered and interactive to boot.


I agree! His articles always have fantastic content!


Why does the first slider with the cube not say what the “one property” the slider controls is? Viscosity? Airspeed?


You're kind of correct on both guesses. You can get that change by changing the viscosity OR the airspeed.

He elaborates later on, but you're changing the Reynolds Number - a calculated value from the velocity, fluid density, viscosity and length. The cool thing about a Reynolds Number is that you get identical (in theory) airflow characteristics for two setups with the same Reynold's Number, even if e.g. the airspeed is different.


I think this is entirely intentional. All articles by Bartosz build up from simple basic principles and avoiding specific technical terms is a good way to onboard viewers of mixed backgrounds without scaring anyone off. Viscosity is actually mentioned for the first time only roughly three quarters into the whole thing.


I think it would benefit by being broken up into modules and a little less simplified, at least naming the things.


Maybe you learn best that way, but that doesn’t mean everybody does


From the HTML:

   <div class="slider_viscosity" id="fdm_hero_sl0">(...)


Also, it says "this substance", which I initially thought referred to the cube as it was just mentioned in the previous sentence. But I guess it's the "fluid".


Pretty impressive. I was curious how they made the whole thing so I went to look at the source for the images. It's mostly in one 10000 line JS file which draws all the graphics onto <canvas> in JS, plus a bunch of WebGL that goes over my head. The code looks like

    function draw_car(ctx, rot) {
    ctx.save();
    let sc = 0.04;
    ctx.scale(sc, -sc);
    ctx.lineWidth \*= 1 / sc;
    ctx.translate(-286, -51);
    ctx.beginPath();
    ctx.moveTo(463.93652, 9.89137);
    ctx.bezierCurveTo(462.12793, 6.72347, 461.22363, 3.5344, 461.22363, 0.32417);
    ctx.bezierCurveTo(447.58911, -1.16177, 434.20691, 2.81333, 434.85754, 5.10777);
    ...
I wonder what their workflow was. Surely all those curves weren't programmed by hand?


Not sure what you are looking at but https://ciechanow.ski/js/airfoil.js is the JS that contains the code for the graphics/visuals. And it is completely readable.

The 2d part though is probably generated.


Yeah, the "draw_car" snippet quoted by parent is from the JS file that you linked.

It looks like maybe an SVG file converted into JS? Do you know if there is some standard tooling that generates this?


SVG paths translate pretty directly to canvas commands. If you have an SVG path parser, it's pretty straightforward to walk it and output the equivalent js.


But for that many paths it seems... tedious. Wondering if they had a better way.


Yeah, that's the file I was looking at. I was wondering how that file got created. By hand, or generated somehow?


I asked him on Twitter about one of his previous articles, he said he writes all that JS by hand.

https://twitter.com/BCiechanowski/status/1522067904522428417


It doesn't look like it's publicly viewable but he has a writeup of his process for patrons at https://www.patreon.com/posts/on-airfoil-99289324


Just a small note, that code looks like Context2d rather than WebGL unless you were looking at something else in the code


Ah, I wasn't clear. There's a bunch of both in the same giant file.


The wings of an airplane in level flight direct air downward with a force equal to the airplane's weight. If one were to build a large scale on the ground, as an airplane flies over it, the scale would register the weight of the airplane. The wings act like a scoop forcing air downward behind the wing. At least that's the way I think about it when I'm out flying around in my Cessna.


Although it is a nice mental model, that's not quite true.

> The wings act like a scoop forcing air downward behind the wing

Only bottom side of the wing acts as a scoop, creating positive pressure. Upper side, in opposite, creates negative pressure which "sucks" the plane into it, creating additional lift.

It surprised me how much lift is coming from the negative pressure - about a half: https://aviation.stackexchange.com/a/16202


Actually, it is quite true. Gravity is exercising on the airplane a force F equal to the weight of the plane, towards the ground. For the airplane to stay at the same height, air needs to exercise a force that is equal and opposite to that of gravity. For an airplane buoyancy is negligible, so the force comes from accelerating enough air towards the ground so that F = M*A when M is the mass of air being accelerated, and A the (average) acceleration.

Notice that this isn't a separate effect from the effect of pressure - it's just a different way of seeing the same effect. The wing is accelerating the air both upwards and downwards, but because the pressure is higher below the wing than it is above it, more air is accelerated down than it is accelerated up - which lifts the airplane, but makes the air go down.


GP was not disputing the redirection of flow or the magnitude of force/air momentum change. They were just saying that not all of this is because of the "scoop" effect from the bottom of the wing: a significant part of the redirection also comes from the low pressure above the wing (at least in practical cases).


Except that negative pressure is not a thing. Air molecules are not grabbing the wings and pulling them up - they are just not pushing down on the top as much as the ones underneath are pushing upwards.


Negative pressure is not a thing, except you just described it.

If you take the difference between the pressures above the wing and below the wing, you get a negative number.

A thing not existing absolutely can still exist relatively.


That’s just a pressure differential, and not what the OP meant by ‘negative pressure’. 100% of the lift force on a wing is attributable to the pressure differential across it, after all.

They (or their stackexchange source at least) are - like the referenced article and as is commonly done in aero engineering - subtracting out ambient pressure as a reference pressure, and then viewing pressure above the wing as ‘negative’ and pressure below as ‘positive’. It’s a convenient choice to make, for various reasons, but it is essentially an arbitrary one.

The problem comes when you then go on, like OP did, to come across statements like “how much lift is coming from the negative pressure - about a half”

Now, since in analyzing the pressure we have subtracted the reference pressure and made a zero point in between the low pressure value above the wing and the high pressure value below it, it actually shouldn’t surprise us at all that ‘about half’ of the lift seems to be attributed to the positive pressure below the wing, and half to the negative pressure above the wing.

This is just saying that half the lift on the wing is attributable to the first half of the pressure differential across the wing, and about half the lift attributable to the other half.

One of the problems of using a relative pressure and thinking about negative air pressure is that it gives the impression that negative air pressure, like positive air pressure, can grow arbitrarily large. It can’t. You can’t have a negative air pressure lower than negative ambient air pressure, because the absolute air pressure cannot go below zero.

But what you’re talking about is a relative pressure differential. We can have an arbitrarily large negative pressure differential because we can have an arbitrarily high pressure on one side of it.


It's not arbitrary: negative gauge pressure above the wing means that (by definition) there is a pressure gradient increasing away from the wing (because the absolute pressure far from the wing is ambient pressure), so the net force on the air there is downward.

> made a zero point between ... shouldn't surprise us

Whether or not you are surprised is immaterial, but it is not guaranteed a priori -- you could get a net upward force with ambient pressure above the wing and positive pressure below or with ambient pressure below the wing and negative pressure above (meaning gauge pressure, relative to the ambient pressure distant from the wing, to be clear). The person who started this thread seemed to be implying that the former was a good mental model, and the person you replied to was just saying that in fact for practical wing designs it is somewhere in between.

FWIW it is very common to talk about positive and negative gauge pressure. Some people may say that without understanding what is going on, but it is a mistake to assume that they don't understand just because they use that language.


That’s my mental model as well. The incompressible fluid-based explanations never made much sense to me


Ya, I was hoping for more nuance related to this. I'm sure the air foils generate lift, but atmospheric pressure at cruising altitude is ~4psi, and the pressure differential across the foil must be only a tiny fraction of that. According to my understanding of Bernoulli's principle, you'd have to quadruple the speed to cut the pressure in half, and I can't imagine the top air traveling that much faster than the bottom air.

Yet a 747 can produce 850000 pounds of lift with only 729000 square inches of wing? Feels like a very incomplete description at best


The pressure differential is what causes the direction change of the flow, pushing the air down. The shape of the wing and the angle of attack cause the pressure differential.


The airfoil shape causes formation of vortex around the wing, which ridiculously changes the relative speeds and pressures involved. At low pressure you compensate with speed, which is squared in lift equation.


... I'm honestly surprised it's possible to get PPL(A) without learning about wing vortices responsible for lift generation.

In order to use "scoop" approach for lift, you need to have either very low wing loading (think paper airplanes) or very high speeds (above transsonic range).


> If one were to build a large scale on the ground, as an airplane flies over it, the scale would register the weight of the airplane

No, it wouldn't.

I think the article does a pretty good job building a more complete understanding than the simplistic "deflection" mental model.


I think what they were saying is that from a pure "Newton's 3rd law" standpoint, if the plane has an upwards force, then the air has a corresponding downward force, which must go somewhere. Yes, it is spread out and complicated and turbulent, etc, but ultimately must balance out.

If we could somehow "draw a box around" the entire plane+air system, then the plane's upward lift will create a corresponding downward force on the box, one way or another.

So, in the broad sense that you push the earth away from you when you jump, the plane also pushes the earth away from it when it flies (mediated by a bunch of fluid dynamics).

Or, classic example: if a (sealed) truck full of birds is jostled so that they start flying, does the truck weigh less? [1]

[1] https://www.youtube.com/watch?v=lVeP6oqH-Qo&t=35s


It's wrong though. A large, hypothetical scale under the plane would not register the weight of the plane as it flies over. And not just because diffusion but that being one of many reasons.


I'm curious to know your reasoning more.

Certainly if we flew the plane very low over the ground, the air pretty directly pushes down on it, and the hypothetical scale would register something. Just look at the grass when a helicopter hovers over it.

As the aircraft flies further up, we'd need a bigger scale to capture the full area affected, and if it's moving there would be increasing lag between the location of the plane and the (large) area where the downward force hits the ground.

Or do you disagree with that? At what point does the scale stop working?

Obviously there would be practical limitations — that force is so spread out that it would be hard to measure. But let's not have practice get in the way of theory (:


The "stone skip" or "deflection" theory of lift is not accurate; https://www.grc.nasa.gov/www/k-12/VirtualAero/BottleRocket/a...

Planes fly through gas, not solid particulate. Gas has intrinsic kinetic energy when energized. Diffusion plays a huge role in all this of course.

The airflow is split at the leading edge. The area of positive pressure is not entirely below or focused under the wing. The top and bottom of the airfoil are both involved in turning the air flow.

The pressure under the airfoil increases a bit, but the pressure above decreases by as much to much more depending(2-3x or more). This hypothetical scale is under the aircraft but much of the lift occurs by decreasing forces on the top surface.

Scales measure weight/mass. Barometers measure changes in atmospheric pressure. So it's not even the tool for the job even if the stone skip theory of lift was accurate.


I think I see what you're saying.

Perhaps my mention of Newton's third law gave you the impression that I was advocating for that "stone skip" theory ­— I assure you I wasn't! Especially as presented on that page, it is obviously junk (:

But surely you agree, broadly, that if birds are flying inside a sealed box, the box still weighs the same amount as if they were standing, right? (modulo some fluctuations)

All of the pressure differentials and whatnot have the net effect that an upward force on the wing results in a downward force elsewhere. The purpose of the scale is to measure that force — like measuring the weight of the box with birds in it.

In the hovering helicopter example, wouldn't you agree that a (large) scale directly under the helicopter will measure a weight corresponding to the helicopter's lift force? Like if I blow directly onto a kitchen scale — it will measure some grams.

Edit: I feel we are kinda re-hashing the Bernoulli/Newton discussion also addressed at the nasa site: https://www.grc.nasa.gov/www/k-12/VirtualAero/BottleRocket/a... — 'So both "Bernoulli" and "Newton" are correct ...'

For reference, the correct Newtonian explanation (flow turning) is also covered: https://www.grc.nasa.gov/www/k-12/VirtualAero/BottleRocket/a...


The simple newtonian deflection model is correct however, As you engineer your deflector to have the least possible drag the airfoil shape naturally falls out.

Actually that is a bit of a lie, the airfoil shape only falls out due to a third implied force that needs to be accounted for. the wing needs to be strong enough to hold itself up. if you had infinitely strong materials the deflector shape that would fall out would be like a slightly bent piece of paper.

A clarification note on fluids: you are deflecting fluids, and everything this implies. just because I say newtonian deflection don't think I mean billiards balls, or if it has to be billiard balls think trillions of them simultaneously



I did not say reflector as implied by that link but deflector, a thing put in the fluidstream to move it somewhere else. airplanes lift because you are moving air down. People get hung up about the convex side of the airfoil but what else is the fluid going to do, stay a vacuum? it is going to move in the way the deflector shaped, adding to(actually providing most of) the downward flow. There is a lot of engineering that goes into it but at the end of the day an airfoil is the shape that moves enough enough air downward with the least drag. The only reason it is a thick teardrop shape is it has to be strong enough to support itself and the airplane. otherwise the ideal shape would be super thin shaped like the upper surface of the wing bending slightly from the cord(aspect directly into the stream) to the trailing edge(a few degrees of slope).


IDK man, what I said was correct; it's wrong. GLHF.


As an aerospace major in college, this has content from several semesters combined into a really well structured, clear and visual explanation of various aspects of flight dynamics. Thank you.


For anyone really interested, this is the authoritative reference for NACA, etc. airfoils.

Theory of Wing Sections by Abbott and von Doenhoff (1959)

https://www.amazon.com/Theory-Wing-Sections-Aeronautical-Eng...


For anyone interested in bio-inspired aerodynamics, there's a fascinating study published in Scientific Reports exploring the aerodynamic efficiency of dragonfly and NACA4412 airfoils in ground effect. This research uses URANS simulations to offer insights into optimizing MAVs' (Micro Air Vehicles) performance close to the ground. https://www.nature.com/articles/s41598-022-23590-2


This looks incredible as usual. What puzzles me, though, is why some people find flying puzzling. At least the kind that we do, ie. helicopters and fixed-wing aircraft. It's easy to accept a fan works: just put your hand there and feel the draft. A wing is just like a linear fan pushing air down. It's completely intuitive to understand for me. The difficulties are just in making it practical and controllable. Conversely, many people don't seem concerned at all with bird or insect flight, which I find a lot harder to understand.


I think many of us were taught in school that airfoil shape was somehow magical -- that the fact that it was bowed more on the top was responsible for the fact that it worked.

This is only partially true, though; a totally flat wing can also support flight. The shaped nature of the wing contributes to its efficiency (and other factors) but do not make other wing shapes incapable of supporting flight.

The reality is that the Wright brothers' innovation was not the airfoil shape or even the lightweight motor. It was the control surfaces, to allow the operator to adjust the plane's attitude on the three axes of rotation, allowing actively stabilized flight.

Paper airplanes and kites demonstrate all the same principles of heavier-than-air flight (the Wright brothers even had a kite version of their airframe they used for testing), despite the fact that they generally do not exhibit shaped airfoils.


The Wrights did use a rudder and "horizontal rudder" on the 1903 Flyer, but they were for some time determined to achieve roll control by warping the wings rather than using control surfaces, and were only forced to adopt ailerons as other pioneers began demonstrating how superior a paradigm that was. So they don't deserve too much credit on that score!


"Control surfaces" was more specific than I intended; what I meant was that their plane allowed them to control all three axes of rotation, and that was the innovation - that they could control pitch, yaw, and roll independently and that allowed them to have active stable flight.

Without those controls, flight is basically impossible, and with them, you could use nearly any airfoil shape (modulo engine power, drag, and stall speed considerations) and achieve heavier-than-air flight.


Ailerons were really only invented when they were (and named in French) because the Wrights were extremely litigious, they sued Curtiss for using ailerons and basically destroyed American aviation for a decade allowing the French a temporary lead. This had an interesting cultural effect of lots of things becoming named in French across aviation (including things like the weather code for mist being "br" for brume to this day).


Because the explanation in school misses something like 90% of the detail replacing it with zero-explanation magical thinking.

For example, yes, the air above the wing moves faster than the air below the wing, and it's related to shape of the airfoil.

However, it has nothing to do with magical "air has longer to travel".

It starts with how combining flows at the trailing edge of the airflow create a vortex which induces an opposite vortex around the wing, which is a bit counter-intuitive (but it has nothing on why swept wings work, which can be summarised for practical aircraft design purposes of "because if we calculate at an angle we get better values and reality is crying in the corner")


> It starts with how combining flows at the trailing edge of the airflow create a vortex which induces an opposite vortex around the wing,

Wait, I was under the impression this Cutta circulation was a computational simplification and the "real" reason were the pressure differences as explained in this submission. What am I missing?


Essentially the work in the article shows the harder to grok, but still half of the whole equation, with only one small mention of an effect that points to the wider environment. Essentially, this is a more close-in view of the airfoil without consideration of the wider flow around.

One comment already mentioned how position of flaps could have visible effect on pressure sensors in front of the plane, and this is slightly mentioned in how the pressure created by front of the air foil has an impact on air "at a distance" from the airfoil.

The vortices created around the airfoil result in significant change of flows, which especially at low speeds provides big chunk of the pressure changes necessary for the creation of lift, with the effect IIRC getting lower as you go faster, with transsonic regime breaking it - because that's when the resulting speeds go beyond speed of sound at given pressure in the air, which in very simplified way means that air can't move towards front of aircraft anymore in those areas, breaking all sorts of flows you depend on at lower speeds.


The whole air has longer to travel thing is obviously hand waving a lot of different properties that are all combining to get better efficiencies. For example, don't forget the coanda effect and its contributions to the shape of a wing. Luckily we can always just return to the navier-stokes equations to help us out.


Growing up I got the "air has longer to travel on the top of the wing than the bottom" explanation, and it always smelled like BS. This is the first explanation of flight aerodynamics that really made sense to me — incredible article as always from this author.


Without their wind tunnel optimized airfoils, the wright flyer wouldn't have flown. Without the controls, it wouldn't have flown. Without the high power to weight ratio motor, it wouldn't have flown. Which was the most critical?


Surely it's a less impressive result that something powered by mains electricity can move the air in a draft than that a multi-hundred-ton aircraft can fly over the highest mountains.

It's the size of the aerodynamic forces and the complexity of the physical mechanisms that create them that many people have trouble with. In particular: intuitions can be pretty wrong, most simplified explanations are wrong under simple experiments, and the problems exhibit scale variance that is unfamiliar (e.g. Reynolds number).

One time I was working on air data computer for a transonic aircraft that could fly up to about M0.95 - during flight test, an air data probe mounted on a nose boom was used to supply impact and static pressures, angle of attack and sideslip etc. for various air data calculations like airspeed and altitude.

I was fascinated that there was a term in the calculation that related to the aircraft flap position - what's happening way out on the trailing edge of the wing actually has a meaningful effect of pressures measured on a boom out the front of the nose during certain regimes of flight.


It's just a matter of scale. What's impressive to me with the big aircraft is that we can organise thousands(?) of people to build something that big. But when it comes to the principle of flying it's just a bigger version of the fan. If you were to say they used the same amount of energy as a fan then that would be impressive. But they don't, they burn tons of fossil fuels. Geese can fly over the highest mountains too and all they eat is grass.


>it's just a bigger version of the fan

I mean, actually, it isn't - that's the whole point about scale variance and Reynolds number and why wings that work for insects are not the wings that work for jumbo jets.


This is an incredible (over)simplification of flying.

We still don't have a very good understanding of turbulence.

Navier Stokes equations still make Aerospace engineers drink.

Yes its stupid simple if you care about the simplest of analogies, but if you try to understand it, there are reasons why 80% of people drop out.


I think you mean incredible oversimplification


Because an airplane doesn't move its wings like a bug or helicopter, and it's wings aren't shaped like fan blades. One might look at a plane and conclude that since the wings and engines are parallel to the ground, it must only move laterally.


Of course it moves its wings. That's what the runway is for.


Rectilinear fan blades are shaped very similarly to aircraft wings. And it does only move laterally until the ailerons are moved away from being parallel with the ground.


Who are these mysterious people you're interacting with who are concerned with (but don't understand) the physics of flight, but who are also not concerned with bird or insect flight?


I guess this is somewhat counter intuitive:

https://www.youtube.com/watch?v=NBsvzMi9-f8

So yeah, fans are puzzling too.


If you watch it very slowly, the paper initially folds under the mouth and then it blows out straight.

I'm guessing the initial puff creates a high pressure area on top of the paper, rolling it downward and back. Them after the puff has pushed the air away, there is now a low pressure zone on top of the paper which lifts it up as the air below is rushing upwards around the sides of the paper.


That seems like the Coanda effect more than anything.


Something that was never clear to me at this level of detail is how a tailwind enables an airplane to move faster. In other words, if the airflow is coming from behind, the lift equation should fall apart and the airplane should fall out of the sky.

https://www.skytough.com/post/tailwinds-make-plane-faster


Our intuitive experience with wind on the ground is wrong. Next time it’s windy outside imagine the entire volume of air stretching out for miles and miles moving across with the wind speed, we’re just standing at the bottom of this vast air ocean. It will blow your mind and you’ll think about wind differently from then on. So with that in mind, once the airplane is in the air, it doesn’t “know” if there’s a headwind or a tailwind at all, unless you have a way to reference the ground somehow (for example, with a GPS) - just like a boat doesn’t “know” it’s carried by a current downstream. If you are still on the ground, it is very possible that the tailwind is strong enough for you to not be able to takeoff in the available runway - but then you would go in the opposite direction or more likely sit the storm out :)


> So with that in mind, once the airplane is in the air, it doesn’t “know” if there’s a headwind or a tailwind at all

This stops being true for quick changes - because there's still the inertia of the aircraft. So if the wind speed changes quickly, the aircraft can't immediately move along with it.

This is why gusts are so dangerous to landing aircraft. A strong gust from behind can cost you all your lift, and a strong gust from the front can temporarily stall your wing.


I’ve noticed this effect while diving. When you’re in a current, you’re basically the same density so you’re moving with the water. In mid-water with poor visibility, this is really freaky, because you have no way of telling in what direction you’re moving, and how fast. If you “forget” your orientation, you can’t really recover it. Thankfully, you always have highly accurate depth gauge, but as for lateral movement, it’s an eerie feeling. You could just pop up anywhere.


Yeah, I wish meteorologists explained this concept better to the general public, since they're basically poised for it. One day I was just wondering where wind _started_ from, and started digging deep into the topic, but essentially we're all just standing on the bottom of a roiling ocean floor that is very sensitive to heat changes from the sun.


Airplanes are always traveling forward relative to the wind, at some angle of attack. Tailwinds don't work by blowing against the airplane's surface and pushing it forward. Since the airplanes are themselves traveling at, say, N kt forward relative to the wind, then if they are inside a 10 kt tailwind, they'll be doing N+10 kt over the ground, if they are inside a 50 kt tailwind, they'll be doing N+50 kt over the ground. If they are inside a 25 kt headwind, the'll be doing N-25 kt over the ground.


The plane is just up in the air moving relative to the air around it, it doesn't care how the air it's in is moving relative to the ground.

A tail wind is just saying that the air is moving in a certain direction with respect to the ground (the same direction the plane is flying). The plane doesn't give a shit about that.


It also changes a lot for sailboats, and even more for faster sailing craft (windsurfers, etc.).

You can feel a much stronger pressure in the sail when moving towards the wind on a fast windsurfer/windfoil as you can do 15-20kts 45deg towards the wind, giving you an apparent wind that is 10-14kts stronger than the true wind.

On the same craft, going away/downwind, you will feel the apparent wind at a similar angle 10-14kts less. In fact, because of the change in drag and forces, you'll probably be going faster and feel even less wind on the downwind leg.

When you turn, this can be a big benefit for going downwind (jibing) as at some point the sail feels zero apparent wind (your motion cancelling out the true wind), feels very light in your hand, and easy to rotate to face the other way. Even knowing the physics of it, the timing and execution is still something that takes a lot of practice...especially on big race gear with a 9.0m2 sail.


Yes that makes sense at a high level, but there must be a point of transition between calm air and a jet stream that makes the wings useless to the airplane for at least a few seconds.


These sudden changes do indeed happen in stormy weather, as adjacent layers of air can move with different velocities relative to the ground (the technical term is “wind shear”). If an airplane climbs or descends through those it will look like your speed (relative to air) is suddenly increased or decreased by some amount and you would have to compensate. It’s also a bigger problem for large, heavy airplanes as you have more work to do to accelerate for a given amount of speed loss.

Jet stream boundary is usually not this sharp, and the airplane would fly much faster than the difference anyway.


It's relevant in practice when landing against head wind. You need to have extra speed to not stall when you enter the slower air near ground.


Do you mean landing with a tailwind? A headwind should allow the plane to create the same amount of lift it needs to avoid stalling at lower ground speeds.


Yes, that’s correct, but the headwind stops being so headwind-y near the ground, so your plane needs to go a bit faster to compensate for the loss of headwind-ness in the seconds before touchdown.


On the flip side you also get ground-effect when you are low to the ground where the high-pressure underneath the wing gets trapped against the ground creating a cushion of pressure increasing lift.


Which can be a bit of a challenge when trying to land, especially for aerodynamically efficient aircraft.


If you took a plane flying in still air and magically, instantaneously replaced all the air around it with a tail wind equal to its velocity then yes the plane would stall and fall out of the sky.

Fortunately that kind of instantaneous change doesn't happen in real life.


GO AROUND, WINDSHEAR AHEAD!


Indeed it does, that's effectively what turbulence is.


Presumably the plane would accelerate as it climbed through the velocity gradient, never falling to a point of negative air-relative airspeed.


Yes, the wings are useless once the air is moving close to the speed of the plane. Thankfully, we have jet engines that help planes move a lot faster than the 100-200 knots that jet streams can reach. They'll still affect the flight but only temporarily.


The bottleneck when it comes to a plane going faster is drag which increases with the square of velocity relative to the air. More drag means the plane has to consume more fuel to stay at its current velocity. So if a plane normally goes 600 mph with no wind then a 100 mph tailwind will allow that plane to go 700 mph relative to the ground to experience the same amount of drag as if it were flying at 600 mph on a day with no wind.


It's like swimming in a stream. Even if you did nothing you would move basically at the rate that the water is moving.

Lift only comes from the interaction of the air and the wing, so if there's zero relative motion then you will fall out of the sky, regardless of if you have a 200 knot groundspeed.

This also means that, if the wind at altitude is above your plane's stall speed, you can hover in place by flying straight into the wind! (example here: https://www.youtube.com/watch?v=n_e6ijREScE)

Similarly, if you are in a packet of air that is moving at 200 knots, the fact that you are moving at 500 knots indicated airspeed does not mean you are flying supersonic from an aerodynamic perspective, despite having a groundspeed of 700 knots.


Yes airplanes have to travel faster (in terms of ground speed) to not stall. This is why head winds are preferred for landings and take offs as it allows ground speed to be lower. But during cruise you want a tailwind to reduce the amount of drag for a given ground speed.


This is a kind of relativity thing.

The only connection a plane has to the universe is the air around it. It simply does not know or care what the ground is doing until the ground is quite close.

A small plane in a very high wind is perfectly happy having a "backward" ground track.

Same thing as if you were trying to swim upstream in a fast river. How fast you move through the water doesn't have anything to do with how fast the water is moving across the land.


It works by reducing the amount of CSS the plane needs to carry.


Both drag and lift depend on the speed of the airplane relative to the wind, not relative to the ground. So, if the maximum efficiency dictates that the airplane should travel at speed X relative to the wind, and the wind is flowing at speed Y in the same direction as the airplane needs to travel, then the airplane, flying at maximum efficiency, will be travelling at speed X+Y relative to the ground.


Planes move through the air, and relative to the air mass.


I really enjoyed reading this, and felt excited when the author promised to explain viscosity at a particle level. But there was just a short presentation about two colliding molecules and I didn't understand the connection to viscosity. It's like a section is missing or something..?

How does viscosity work?


viscosity has very interesting units - stress (force / area) divided by rate (1 / time). viscosity is measured (a field known as rheology) by, in some way, moving a thing through a fluid at increasingly fast accelerations, or equivalently, at increasingly high frequencies. that is, imagine moving your hand back and forth in a fluid - the faster you do so (the number of back and forth motions per second), the more resistance you will feel from the fluid. for newtonian fluids, the resistance you feel (measured in force / area, ie the area of your hand), is proportional to the frequency of your hand moving back and forth in the liquid, so, the graph is a line. non newtonion fluids do not have a linear relationship between shear stress and shear rate. air is also a fluid - all gasses are, and thus possess rheological properties. air, however, at stp, is essentially an ideal gas, that is, it is non-interactive, and thus, has 0 viscosity. the point here, is that viscosity is a consequence of the interactions of particles. as gases become denser, their viscosity increases. liquids, for comparison, is ~1000x as dense as air. the details of how molecular interactions lead to viscosity is actually quite complicated.


Thank you. I seem to have trouble using rate as a concept, especially dividing by it :) But I think I get it when I add a virtual distance into what you are saying.

You are saying (force / area) / (1 / time). I add two distances that cancel out: (distance * force / area) / (distance * 1 / time) and get (energy / area) / speed, which is energy used per area and speed. I can feel that, and it seems to be what you are saying, right?


yes, this a very reasonable though inexact interpretation.


Not really a full answer for you, but one thing that this page clarified for me:

I had generally previously thought of viscosity as "how slow" a fluid is. High viscosity means high "thickness," which means it flows slowly (like molasses vs. water).

But as presented on this page, viscosity is actually a measure of "how fast" — how fast the effects on one molecule can spread out from there to neighboring molecules. Perhaps you could think of sounds waves moving through a substance — a "thick" substance like solid metal propagates those waves quickly (on a molecular level), while with a "thin" substance like air it's much slower. In the more precise language from the article: "viscosity controls the diffusion of momentum..."

So, because this diffusion happens quickly in a high-viscosity situation, little whorls of turbulence are inhibited, because the forces governing those whorls get spread out/diffused quickly.

Perhaps you missed the part of the article talking about diffusion, or did not see the connection? The link between that and viscosity was not immediately apparent to me, either.


Thank you, this feels helpful.

Though I don't think I missed a part of the article, I feel more like the author did ;)

What I still don't get is what the difference between high and low viscosity looks like on a particle level. I don't understand why he introduced the collision between two molecules and then never explained that.. :)


I'm taking a sailing class, and learned sails in addition to keels utilize this concept.


The same author has a great similar piece on mechanical watches. Love the format!

https://ciechanow.ski/mechanical-watch/


What an astounding amount of work must have gone into that post...


This guy's whole corpus reminds me of Nevil Shute's _Trustee from the Toolroom_: quiet, fastidious, and expert work appreciated worldwide.


I'm surprised hang gliding doesn't get more attention. It's the closest human activity to flying like a bird


That page made my laptop fan start roaring.


You're probably lacking webgl HW accleration.

https://webglreport.com/ , or check about:support if you're on firefox.


Thanks. Yes, it's Firefox, but apparently it supports WebGL 1 & 2.


yes but is the rendering performed in GPU, or in software is the question


How do I tell? I tried about:support and webglreport.com.



So I disabled "Use recommended performance settings" and "Use hardware acceleration" was enabled.

The fan still roars.


Solved (I think): install firmware-amd-graphics.


If you find this interesting, I would also recommend https://ciechanow.ski/bicycle/. It is by the same author.


An important question before I read something that long (or anything, really): Who is Bartosz Ciechanowski?


Not a single mention of Bernoulli. Bravo!


I expected to see a mention of the Joukowski airfoil (https://complex-analysis.com/content/joukowsky_airfoil.html). I guess there was plenty of other content to cover though.


Amazing work as always.

Is it just me who sees diagonal lines on the first interactive graphic? Are they a bug? They appear after a little while, fade out, then reappear

https://gist.github.com/RobinL/aa4a5e14d35e61e46c1e99c8198d0...


As part of building my own truck-top camper, I got into researching aerodynamics of vehicles in order to try to reduce loss of fuel efficiency. The most interesting ideas I found were that aerodynamics don't matter much on most vehicles until they pick up significant speed.

Most automobiles are pretty heavy, so the engine has to do significant work just to get it to move. At a certain point, the vehicle can change gears to get the engine to do less work and use less fuel. But around the same time, the force of the air is increasing. By the time an automobile goes over about 50mph, the air forces are getting increasingly strong, and the engine has to work harder to keep the vehicle moving. At this point, beginning to lower the air's coefficient of drag on the vehicle will lessen the work the engine needs to do to keep the car moving at speed. So you can optimize the design of the vehicle's exterior to reduce the drag coefficient, which will reduce things like flow separation and turbulence, creating fewer rear pressure zones and causing less drag.

So you might wonder, why aren't more cars teardrop-shaped like the airfoil? The answer is, it depends. Most people want something that looks good more than they want efficient operation at speed. But sometimes having more drag actually helps. For example, the Lotus Elise: while it is smaller and looks more sleek than a Tesla Model 3, it actually has a much worse drag coefficient than a Tesla Model 3. The Lotus has way more force acting against it at speed than the Tesla. But the Lotus is a sports car, and sports cars benefit greatly from increased traction, and you can get more of that traction by increasing the downforce on the car. So the Lotus's design sacrifices top-speed drag coefficiency in order to gain some downforce which helps traction when cornering at speed.

What about pickup trucks? Even though modern pickups actually have lots of subtle design changes to improve drag coefficient, they all tend to have open beds, which is terrible for drag. It creates this giant messy turbulent pressure area in the bed which drags on the tailgate and the rest of the car. By adding a truck topper, the drag is significantly reduced, but you don't see most trucks driving around with a topper on. But trucks naturally have worse gas mileage, so nobody really thinks twice about the aerodynamics.

(To be fair, the air's impact on gas mileage is minimal unless you're going quite fast. But for trucks with extremely bad gas mileage, like 18-wheelers, it makes much more difference. That's why they often have airfoils on the front of the truck, gaps between cab and container closed, and skirts to reduce drag from the undercarriage. Strangely though, the biggest improvement to reduce drag coefficient actually comes from modern European big-rigs whose containers are actually tapered like a teardrop. The rear of the vehicle's shape makes the most difference to how severe flow separation is, and thus how big of a pressure area develops, pulling on the rear of the vehicle. If we wanted to make trucking more fuel efficient globally we'd change the shape of the containers to be more like teardrops, but that would make handling and shipping them much more awkward)

You'll usually only see these effects on automobiles at higher speeds, due to the vehicle needing to overcome gravity before the air forces build up. Lighter vehicles (say, bicycles) with less impact on them from gravity will be impacted earlier (at lower speeds) by the force of the air, so optimizing drag coefficient is much more important, which is why bicycle racers have to put so much into aerodynamics at significantly lower speeds than an automobile. Interestingly, the drag coefficient on a bicycle and rider is actually equivalent to that of a small car.


Obligatory XKCD comic (also with the name "Airfoil"):

https://xkcd.com/803/


Ciechanowski is likely the best content producer we know, absolutely fascinating reads. Imagine having such a person as a teacher - he could probably excite students about any scientific topic.

I'd love to spend my time working on such articles when I'm retired :)


Without doubt. Question is why so many people who try to teach approach it like it’s 1878.


In what ways do so many people who try to teach approach it like it’s 1878?


In the ways that

* still the classroom is what it looked like in 1878

* still a lot of learning is about remembering useless facts

* still there are many places where the curriculum is a static thing which does not reflect the needs of youngsters but their grannies

* still the exams are solo even though later in life one works in a group and it was shown that in a group one manifesta better

* many places still don’t teach topics such as personal finance, digital hygiene, business management, soft skills, etc as universal mandatory classes even though the world we live in requires these skills on a daily basis

Neil Postman makes in depth analysis in his books, and particularly The End of Education.

I can also go for ages with examples from my practice and what I see, and you can downvote me as much as U like. It doesn’t chang the fact that organised schooling is a byproduct of Industrial Revolution and it was designed to cater to its needs. Not the needs of the ages we live in… like so many things tbh which absolutely do not make sense to a teenager at this day.


Wake up babe, new Bartosz Ciechanowski just dropped


LOL! Yeah, so the next should be at least in 2025-Q1.


This man to me is a modern day Da Vinci of the Web


It's truly impressive. the amount of time and dedication its uncanny.


The javascript code is not minified and is easy to follow. Beautifully done.

https://ciechanow.ski/js/base.js

https://ciechanow.ski/js/airfoil.js


I've never used webgl -- what's its advantage over normal 2d canvas drawing for a project like this?


Speed. Drawing thousands of objects (such as the blades of grass or air molecules) with 2D canvas will be very slow. WebGL allows all drawing to be offloaded to the GPU, which can draw thousands of objects in parallel given a single CPU command. 2D canvas also doesn't provide any 3D primitives (as the name would suggest), while WebGL natively supports rasterizing 3D triangles with perspective correct texture mapping and z-buffering.

The downside of WebGL is its complexity, but there are many libraries to help with that.


Blows my mind that my computer isn't even turning on its fans considering how many shaders are running in this thing.


I think they pause when they scroll out of view?


They’re simple shaders and it’s amazing what a computer can do billions of times per second.


Yeah, I mean come on, we're not rendering something hard and expensive to compute, like a React website that has to list items in a shopping cart.


Personally all the fluid simulation shaders I've written usually makes my fan go off, and I'm counting a few of those here so that's impressive in my eyes.


Yeah. It's impressive to my eyes as well. I was just trying to make a joke about how normal websites need 100% of your CPU to render some text and images, and here's this guy doing multiple fluid simulations on a web page written in custom WebGL and it runs on a potato.


I'm on Ubuntu and Edge (v122.0.2365.59) could barely render the page. Chrome (v122.0.6261.94) worked just fine, though. I don't know enough about browsers and GPUs to debug why that is, but I checked (edge|chrome)://gpu and nothing stood out as appreciably different.

Edit: interestingly, it seems to only be the first animation. If I scroll it out of view the others all seem to render fine.


His work is always a good reminder the open web is still an amazing place!


I thought this was going to be about pipeline workflows... but then I saw it was Bartosz!

I know what I'll be spending a stupid amount of time reading today :)


I swear Bartosz’ posts deserve to be pinned to the top of hacker news on release day at this point.


If you like his work, here's his patreon: https://www.patreon.com/ciechanowski/


If you'd like to see more of their work, ranked by what HN thought was most interesting: https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...


You can also see all-time top posts: https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...

The mechanical watch post is 6th on the list


His mechanical watch internals page is also equally amazing.


It is currently the 6th most popular post on HN:

https://news.ycombinator.com/item?id=31261533


Imagine a world in which all education was this level of quality.

Imagine how much more you'd know, be able to do and understand.

I really wish good education was valued more highly in society.


Yes, that. Call it enhanced learning? For instance, add Dan Carlin - Hardcore History podcasts for your history lectures. If everyone listened to those podcasts, then all you would need is a good teacher/professor to discuss what you learned - and there is 'so much' learned from any one of his episodes.


While this is good quality, this is not replacement for real education. Real education involves sweat and hard work, not just consumption. This is somewhere between education and entertainment.


What's missing for it to be real education? Genuine question :)


>Real education involves sweat and hard work, not just consumption.

Anything that fits the above as was stated :)


What? A ciechanow.ski airfoil explainer? But I have things to do today!


It is too late for us to be saved, my friend, way too late...


This is the future of education. Very approachable and seems to target the most common denominator of knowledgeable people that out there.

I wonder will there be articles in the future with more math and code snippets?


The future of education is as an entertainment. There will be no need to educate oneself in the AI future (except for merely egotistical reasons?).


Absolutely beautiful article and presentation


Dare I ask for the code?


It's right there. unminified and unobfuscated. just click save


It's a bit sad this isn't the norn for education articles (and mostly everything else too).

Bartosz's dedication and craftsmanship is really inspiring.


Wow, I never realized this detail. What a wonderful thing.


You’ve been moving through life like an idiot


IIRC you can view the source and it's all custom WebGL available for viewing.


[flagged]


I suggest reading the article.

It goes into great lengths to show how lift is generated with symmetric airfoils or even flat planes and that the asymmetric airfoil is for efficiency and conditions.

It does not invoke any kind of magic. Even the part you quoted doesn't. Indeed they even animate a flat plane at different angles of attack showing lift.


I don't read that sentence to mean what you do. The article is indeed "looking at the forces generated by the flow of air around the aircraft's wings" and it's definitely focusing "on the cross section of those wings" which is a shape known as an "airfoil."

Later on TFA says "the shape and the orientation of the airfoil helps airplanes remain airborne" which is closer to your criticism, but still true; a shape that generates more drag or less lift in the equivalent airflow would not help airplanes remain airborne.

Maybe if TFA included a simulation of a rectilinear wing and showed how it stalled at very low angles of attack, that would improve things, but I find it to be "just fine" as an introduction to lift.


    > It'd be like clicking on an article that purported to explain "How computer programs work" .... and then proceeded to describe a debugger.
iirc the article is called “Airfoil”, not “Rectilinear Box”.


To be nitpicky: I wasn't irked by the article's title, but by the framing in the lede.

To be constructively nitpicky: a box used as a wing is absolutely an "airfoil" inasmuch as the term has any meaning. It's not the shape being "special" that makes a wing work, it's the shape that the airflow around it takes, which is to first approximation just a function of its "tilt" along its major axis relative to the flow direction. The business about shape is all just optimization, not what you want to describe if you want to know how an airplane flies.


Very amusing, enjoyed the constructive nitpick!


Indeed, with sufficient thrust your wings can just be flat plates with a small angle between the plane of the plates and the direction of travel. Airfoils are first about reducing drag, second about stall speed and angles.

But airplanes are quite constrained things, if you have a bad attitude and refuse to use airfoils many planes wouldn't even get off the ground and all the rest would have abysmal performance. Sort of like how with imagination everything is a hammer, but this being technically true doesn't mean that you shouldn't really use a hammer when one is called for.


Oh, thanks todsacerdoti. I really need such thing. Actually, I love reading contents which related to Planes, Wings and the base reasons of their flying. It shows that you and your posts are going to be the best items for my hobbies in the future. Thank you.


Things fly because of thrust, not wings. Rockets, missiles, don't have wings and yet they still fly even longer distances. Shut the engines of a 747 and it will fall like a rock, no matter how perfect the airfoils.

Bernoulli and Coanda are important but without thrust/velocity there is no lift


A 747 won’t fall like a rock. It will glide.


You can't say that without mentioning https://en.wikipedia.org/wiki/British_Airways_Flight_009

Where the pilot said what is probably the most British thing ever said in history:

> Ladies and gentlemen, this is your captain speaking. We have a small problem.

> All four engines have stopped. We are doing our damnedest to get them going

> again. I trust you are not in too much distress.


Just in case folks don't click on the link, it was a safe landing and everyone survived. It turns out modern planes can glide quite well.


This is a very important point.

The key difference is that a rock (like any ballistic projectile) accelerates until terminal velocity. In contrast, a 747 (like any airplane) descends with a constant vertical velocity when they lose power.


The engineers at Boeing are going to be really embarrassed once someone tells them they don't actually need to put wings on their aircraft...


not to mention authors of rfc-1925 as well :o)




Consider applying for YC's W25 batch! Applications are open till Nov 12.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: