Hey all, I removed a part of the video that had some misinformation, hence the "jump" from one section to another. I made a large error in what I was conveying, so here's a correction by viewer Jay Raut: From what I understand (and don't quote me, its been a while since I've dealt with fluid dynamics), the problem with the Navier-Stokes equations is the issue of them being ill-conditioned. By that I mean that a small change in the input does not result in a small change in the outcome. This is important since with any system, a small input change should always yield a small output change, otherwise the reliability of the solver is questionable (the results should be reproducible, and near infinitesimal changes should not result in drastically different answers). Now while the the equations are basically glorified F=ma equations, which means that they are most likely the correct DE that describe the underlying physics, the problem lies in the fact that we simply don't understand or appreciate them enough. Also, remember that the real underlying physics is much more complicated. We can break down the problem to its core where we consider all the fundamental forces of the universe and the quantum effects between each particle in the fluid. But, this is meaningless because we want a meaningful compressed description of the physics, similar to how Newton's laws of gravity are a simpler version of Einstein's. I've solved the Navier-Stokes equations by hand in undergraduate classes for simple problems, and in these cases the equations are very well behaved. The solving process is actually very logical to the point where you realise that all you are doing is Solving F=ma. The problem comes down to turbulence, and the fact that the simple Navier-Stokes model do not capture this phenomenon at all. There have been very complicated proposals to the NS equations which take turbulence into account, but these are loosely based on analytical physics and more empirical solutions. Introducing this does not only create a more accurate solution, but employing some numerical trick also make the solution very stable. Also, there is also the problem of the DE itself. Its not simple to solve, and the numerical methods we usually employ to obtain approximation, are exactly that: approximations. So if you read the problem statement more carefully, you will realise that there is no straight forward problem that has to be solved. It's like the people didn't know what to set as the problem itself, which has become the problem. To essentially solve the millennium problem, you would need to come up with some form of proof that the NS equations are truly the underlying physics of a fluid (or not). Remember I mentioned the problem of ill-conditioning? Well even if that is true, that does not mean that the NS is BS, and the turbulence modelling tricks can make the solution very stable. However, these tricks are sometimes based on nothing more than: 'it works'. This is not progressive work and that is what the millennium prize tries to address. So answering the question in terms of your words, we don't know if the solution (real) is smooth. because of which we don't know if using tricks to make our modeled solutions smooth is the correct thing to do to obtain meaningful answers either. And upon finding out whether or not it is, we'd also like to know why? Essentially: solve turbulence, because nobody knows wtf is going on. A second mistake is that isothermal refers to no loss or gain in TEMPERATURE and not heat. Sorry about that, I definitely got a bit confused when typing up the script. I'm considering making a follow-up video as to what was wrong with the video and explain what we are actually solving.
@@UnfinishedEngineer lets say a gaz is compressed by a piston, the temperature of the gaz will increase as we have a higher particle agitation due to high pressure, every variation of temperature is concerved as we dont have any heat transfer with anything. That is adiabatic. If we had colden down the gaz by letting out some of the heat in order to conserve the initial temperature, we would have a constant temperature variation yes but that is because we did a heat tranfer. That is isotherme. You can have an adiabatic isotherm process.
Iso means ‘equal’ like isometric means ‘equal measurements’. So isothermal means ‘equal temperature/heat’ Just pointing this out because I literally only found out recently what iso means and it was driving me crazy beforehand trying to remember the meaning of these names.
Im a med student. Wish i could understand maths easily. Seems much more interesting than human biology. It's very hard for me to grasp these concepts but im not giving up.
@@HexoseMonophosphateCunt Oh man, you must have a lot of free time. I am considering to let go med school to study physics. But ,regardless of what happens, doctors are always welcome. Good luck.
@@everab1209 No man i don't really have lot of free time it's because of covid im stuck at home and have more time, So instead of watching netflix and stuff i prefer learning new things. Thanks though :)
We have to face our karma curve at some point so divergent is something like diversity but we followed the same space so we got equilibrium at some point
Is this something I can argue I need Threadripper for my desktop pc to calculate these? So hard to find any game to actually do demanding calculation, other than synthetic Pi calculation.
Some already pointed out mistakes, some key information left out, but overall a nice video. Having tried myself, I know how difficult it is to make videos like these with Manim, so congrats. Also, nice to see more people doing videos on math subjects.
@@brijeshpr6543 It's applicable for any flow, both laminar and turbulent, but it's complicated, there's a lot of practical limitations regarding numerical simulations. Turbulent flows often require a very refined mesh for numerical analysis or some sort of turbulence modelling that are usually not derived from first principles. Computational Fluid Dynamics is a very interesting, but very complicated, area of study.
@@brijeshpr6543 for every type of flow, laminar, turbulent, compressible, incompressible, steady, unsteady. Just the form changes. For example, in this video mass equation is simplified to incompressible flow, and he pointed it out.
That is not the definition of smoothness, smoothness means that it is infinitely differentiable. (Whatever that means) It comes to the study of functions on smooth manifolds, hence smooth functions. The pendulum, for example, I’m not sure that it’s solution has a closed form, but Banach Fixed point theorem assures us that there is a solution!! And it is smooth!!! Now, you can ask then, what would it mean to not be smooth? Well for example the absolute value is not smooth since it is not differentiable at 0. But more than that, experiments on turbulence have shown that turbulence in fluids looks like a fractal!!! And let me tell you, fractals are not smooth in general!! In my opinion turbulence shows us that there is a loss of structure (again, whatever that means).
Thanks for your comment Jose! I was typing up a correction to this at the same time as I saw you're comment. I got confused with a few concepts when typing up the script. I should pay more attention and run it by a few people next time.
Now since you read my comment, I hope you read this one too. Great video, you have a lot of talent and I encourage and celebrate it! So congratulations, and please keep doing videos!
vcubingx I am just a graduate student, but if you can contact experts to check the drafts for your videos, it may help to rise the quality of your work even higher! Again great job and thank you for your excellent effort! You can contact me and I can answer your questions if I can or even better, get you directly with the great professors from my university.
These are not the Navier-Stokes equations but rather the initial startup of Hagen-Poiseuille equation. You have forgotten the nonlinear convective acceleration term u⦁∇u on the left hand side, which is what this price is all about in the first place. This term is responsible for turbulence and the white water you’re referring at in the beginning of this video. It should be like this: ρ(∂u/∂t + u⦁∇u) = ∇p + μ∆u + F Or with material derivative ρDu = ∇p + μ∆u + F Or more commonly ∂u/∂t + u⦁∇u = ∇p/ρ + ν∆u + F/ρ Where ν = μ/ρ is the kinematic viscosity. It’s a great video though. Time consuming or not, I would seriously change that, because significantly different equations, more than million dollars to say at least.
When they say prove the solutions are smooth, does it mean that the solutions are smooth but we can’t prove it? As you said we can’t predict weather too many days ahead, so that means the solutions are chaotic but we haven’t proven that either? Can chaotic solutions be smooth?
From what I understand (and don't quote me, its been a while since I've dealt with fluid dynamics), the problem with the Navier-Stokes equations is the issue of them being ill-conditioned. By that I mean that a small change in the input does not result in a small change in the outcome. This is important since with any system, a small input change should always yield a small output change, otherwise the reliability of the solver is questionable (the results should be reproducible, and near infinitesimal changes should not result in drastically different answers). Now while the the equations are basically glorified F=ma equations, which means that they are most likely the correct DE that describe the underlying physics, the problem lies in the fact that we simply don't understand or appreciate them enough. Also, remember that the real underlying physics is much more complicated. We can break down the problem to its core where we consider all the fundamental forces of the universe and the quantum effects between each particle in the fluid. But, this is meaningless because we want a meaningful compressed description of the physics, similar to how Newton's laws of gravity are a simpler version of Einstein's. I've solved the Navier-Stokes equations by hand in undergraduate classes for simple problems, and in these cases the equations are very well behaved. The solving process is actually very logical to the point where you realise that all you are doing is Solving F=ma. The problem comes down to turbulence, and the fact that the simple Navier-Stokes model do not capture this phenomenon at all. There have been very complicated proposals to the NS equations which take turbulence into account, but these are loosely based on analytical physics and more empirical solutions. Introducing this does not only create a more accurate solution, but employing some numerical trick also make the solution very stable. Also, there is also the problem of the DE itself. Its not simple to solve, and the numerical methods we usually employ to obtain approximation, are exactly that: approximations. So if you read the problem statement more carefully, you will realise that there is no straight forward problem that has to be solved. It's like the people didn't know what to set as the problem itself, which has become the problem. To essentially solve the millennium problem, you would need to come up with some form of proof that the NS equations are truly the underlying physics of a fluid (or not). Remember I mentioned the problem of ill-conditioning? Well even if that is true, that does not mean that the NS is BS, and the turbulence modelling tricks can make the solution very stable. However, these tricks are sometimes based on nothing more than: 'it works'. This is not progressive work and that is what the millennium prize tries to address. So answering the question in terms of your words, we don't know if the solution (real) is smooth. because of which we don't know if using tricks to make our modeled solutions smooth is the correct thing to do to obtain meaningful answers either. And upon finding out whether or not it is, we'd also like to know why? Essentially: solve turbulence, because nobody knows wtf is going on. I think that last paragraph addresses the question you had about chaos?
I'm pretty sure that's the chaos theory you just described, this on the other hand shows small changes do add up but don't drastically change the outcome. Please correct me if ive misunderstood
The million dollar question is why small changes don't result in drastic outcomes overtime. I think it might have something to do with the correlation between the area said newtonion fluids are operating in.
Very nice one! I have derived the whole NSE as well as the Mass & Energy conservation on my channel to actually grasp the concept of where these equations come from a bit better. You did a great job in explaining the main ideas and problems under 10 minutes, props! :D
When looking at Navier-Stokes the fundamental properties you are looking at are bulk properties and are impossible to define as a individual atoms. The infinitesimals are assuming a continuous fluid where there are no such things as particles. Think of density in the context of a particle, outside of the arbitrary area that defines that particle the density would be 0 and thus the system wouldn't be continuous. Rarefied gas dynamics is the feild of fluid mechanics where a gas is treated as a random assortment of molecules. And uses a variety of methods to figure out fluid flow when molecules are so far apart these bulk properties break down.
2:05 Correct me if I’m wrong but an isothermal proces just means that the temperature remainins constant, not that there is no exchange in heat. In fact an isothermal process means there is no change in internal energy, which through the 1st law of thermodynamics entails that the work done by the system is equal to the heat gained by the system (I believe that was the correct phrasing of the first law given the change in internal energy is 0). So if there is work being done at a constant temperature there must be heat gained or lost.
The equations you are showing represent the incompressible Navier Stokes equations, where flow density is assumed constant (Mach < 0.3). This is already a great simplification of the physics and this subset of the equations will not apply to flow over commercial airplanes (Mach > 0.3) and certainly not to rockets (Mach > 1). The full set is comprised of 5 PDE's, conservation of mass (1), conservation of momentum (3), and conservation of energy (1). Solving these equations numerically by marching them in time from an initial flow condition is relatively easy and straightforward, yet it requires significant computing power.
2:27 We are not describing the behavior of individual molecules of fluid through Navier Stokes equation. In fact, the velocity of individual molecules can be much higher than the flow velocity. Kinetics theory of fluids deals with that topic. In deriving the Navier Stokes equation, we rather treat treat the fluid to be a continuum.
In a real fluid divergence is not zero because you can probably imagine how if you compress it all into the center, the invisible particles WILL bunch up in the center, meaning that there is more mass entering the center area than leaving it. It all makes sense!
I've been working on this project since quarantine started and have made so much progress, so this video came a little bit later... but I was more interested in CFD and calculating things through code. Luckily, after soo many hours put into research, learning all this calculus stuff (currently in 10th grade so I had barely any experience with PDEs lmao) I finally got some C# code working with a Windows Form that allows me to specify the initial velocity, pressures for each cell and can tell me the next frame. Personally a great accomplishment. Something Ill definitely be putting on my college app for my projects during quarantine haha Thanks for making the video!
hmm - so the goal is to prove that there are solutions to the navier-stokes equations that are non-chaotic? because i've heard a different definition of smooth: not as "opposite of chaotic" but rather as "infinitely often differentiable". or are these two things related?
You’re right - smoothness refers to infinitely differentiable, not sensitivity to initial conditions. In fact, we already know that the NS equation is chaotic and exhibits wildly different solutions for arbitrarily small changes in the initial conditions, so that wouldn’t be a very interesting million dollar problem! To win the million dollars, you need to show existence and uniqueness of a smooth solution, or show that a solution to the NS can breakdown (diverge to infinity).
@@talrefae97 yes, i know that the solutions of NS equations are chaotic in general - but that (by itself) does not necessarily rule out the existence of non-chaotic ones. so, i thought, maybe the quest is to hunt these down, if they exist. like, for example - as happened with the 3-body problem: ruclips.net/video/et7XvBenEo8/видео.htmlm16s yes, it's an ODE - but maybe something similar may happen for PDEs as well? anyway, thanks for clarification. or maybe i'm misunderstanding the sense of chaotic here - if it means sensitive to initial conditions - it could very well be that these periodic orbits of the 3-body problem also qualify as chaotic, despite being perfectly periodic. ...i guess that would depend on whether these periodic orbits are attractors ot repellors. edit: ha! here's more about that! ruclips.net/video/eqSPvyaxMI8/видео.html so, some of these orbits seem to be indeed attractors (that's the same thing what he means with "stable", right?)
@@MusicEngineeer This is not the same as what needs to be done for Navier-Stokes. With ODEs we have standard existence and uniqueness theorems which tell us any initial condition belonging to an appropriate set leads to solutions that are differentiable in time, at least locally. The problem with PDEs is that the set of initial conditions becomes spaces of functions (as opposed to vectors for ODEs) and so you have to prove that for any initial condition from an appropriate choice of these spaces (typically a Hilbert space) leads to a solution of the Navier-Stokes equation which 1) exists for all time and 2) remains in the space that the initial conditions belong to. It's the second that is the hard part since the definition of the space requires a bound on the spatial derivative, which is currently what cannot be proven. To get more technical, what can potentially happen is that you have an initial condition that leads to a solution which exists and is bounded for all time, but its derivative blows up at a certain point in time. It doesn't have to blow up as in go to infinity, but blow up in the sense that its norm with respect to some space of functions goes to infinity. That is, the derivative "leaves" the space you started in. Hence, to prove the millenium problem you need to show one of the following holds for the Navier-Stokes equations: 1) All appropriate initial conditions lead to solutions that are bounded and have bounded spatial derivatives or 2) find an initial condition which leads to a solution for which the derivative blows up.
The goal is basically to show that for any given initial conditions there is a single set of evolution functions, which also happen to be smooth functions (differentiable) This related to chaos since numerical methods are about approximating a solution, and a solution needs to exist for this to be the case. And this is not at all trivial. There are plenty of PDE that are proven to not be solvable for arbitrary initial conditions
Div(u) = 0 translate the volume conservation. You can talk about mass conservation only if the density is constant in time and space. Which van you have if you consider the fluid is both incompressible and homogeneous, the latter not being specified in the video
My guy, you did an excellent coverage in this very hard topic, but I don't want to be that guy, but here we go. At 2:06 Isothermal is when the temperature stays constant, but Adiabatic is where there is no loss or gain of heat. but CMIIW
3:50 Div u is part of the continuity equation, not the Navier-Stokes - simple Wikipedia would tell you that. Navier-Stokes speaks to momentum conservation.
Not the point of the video but in the past four years of engineering school using div , this is the first time someone actually explained what exactly is the divergence operator and what it stands for lol
Brilliant! You should get paid by the ministry of education for that! All faculty teachers should use your videos to teach their students (same for 3blue) Cheers, A physics student
Good video, but I feel compelled to point out that in your explanation of a newtonian fluid is, in a strict sense, untrue although I think you get the right message across. Viscosity is an intrinsic property of the fluid. In other words, the viscosity of ketchup doesn't change regardless of whether it is in motion or motionless. What you really meant to describe was the change in the viscous stresses. Again, this probably doesn't matter for the sake of what you are trying to point out, but is definitely important for someone trying to learn these things in more detail.
And viscosity must not be a scalar, it is a 2nd rank tensor (a matrix). In Newtonian fluids it just happens to be a constant multiple of the identity matrix, so we sometimes think of it as a scalar.
yess maybe you could explain Hodge conjecture using simple geometric analogies about determining all shapes (~= homology classes) of algebraic varieties.
So this question occurs to me. is the source or cause or friction in a fluid always the same thing? Just thinking about honey, what are the properties of that substance which cause it to have friction? I know that the friction of honey can be reduced by adding in energy (heat), but I image that there are in fact liquid systems which will not respond to heat that way, or at least I know that some liquids have less friction based on something that doesn't have to do with heat, like motor oil, which heat affects, yes, but at room temperature it is less viscous than honey, so that has to do with something about its molecular make up, not how much heat is in it. (I think) I only ask, because I question the assumption that friction can be thought of as a universal thing, even if the effect is the same, the cause is different, and it feels like that needs to be factored into the equation, or like different liquids would need a unique equation for their properties.
Not per se. Numerical solutions are not great (That's kinda the point, they are very chaotic), and they are for incompresible fluid, which is not ideal in simulations. So blender and so on probably use FLIP or other numerical methods, same principle, but very different concepts!
Sorry but one of my pet-peeves for the NS is when people say "We can replace mass with density". That's NOT what's happening! What you're doing is dividing BOTH sides of the equation by volume, so you're replacing F = m a with F / v = m a / v. As you say, m / v = density. But what can we say about F / v? Well, F / Area = Pressure. So F / v = Pressure / Length. Hence, your F / v side features terms such as "Del p" (where p = pressure) and "rho g". Note that "rho g" is equal to a [ Mass ] / ( [ Length ]^2 [ Time ]^2 ), i.e. the same dimension as Pressure / Length!
Maybe answered elsewhere, but in the intro you say that the NSE can model any fluid...including air, but at 1:52 you say it depends on the fluid being Incompressible. Air is a gas, and near enough under normal conditions, to be an ideal gas - which by definition is NOT incompressible. Its volume always changes directly with its pressure according to PV=nRT, and, if T is allowed to remain constant, the volume changes inversely proportional to its pressure ( V = Const/P). How can both conditions be true? (And no. Air is not "effectively incompressible below Mach 0.3" - when you pump up your tyre, its volume changes with pressure - and nothing is going remotely close to Mach 0.3!)
At 4:45, the second Navier stokes equation has the term F as the external force term. I’ve watched videos from Numberphile and searched up the equation but I keep finding the external force term being either ‘Fp’ (just take the p as rho, density’) or ‘gp’, etc. could someone please explain to me which is the actual right one? Cuz if you jsut divide F by V, you will get density x acceleration in reality.
![The most beautiful equation in math, explained visually [Euler’s Formula]](http://i.ytimg.com/vi/f8CXG7dS-D0/mqdefault.jpg)
Hey all, I removed a part of the video that had some misinformation, hence the "jump" from one section to another. I made a large error in what I was conveying, so here's a correction by viewer Jay Raut:
From what I understand (and don't quote me, its been a while since I've dealt with fluid dynamics), the problem with the Navier-Stokes equations is the issue of them being ill-conditioned. By that I mean that a small change in the input does not result in a small change in the outcome. This is important since with any system, a small input change should always yield a small output change, otherwise the reliability of the solver is questionable (the results should be reproducible, and near infinitesimal changes should not result in drastically different answers).
Now while the the equations are basically glorified F=ma equations, which means that they are most likely the correct DE that describe the underlying physics, the problem lies in the fact that we simply don't understand or appreciate them enough. Also, remember that the real underlying physics is much more complicated. We can break down the problem to its core where we consider all the fundamental forces of the universe and the quantum effects between each particle in the fluid. But, this is meaningless because we want a meaningful compressed description of the physics, similar to how Newton's laws of gravity are a simpler version of Einstein's.
I've solved the Navier-Stokes equations by hand in undergraduate classes for simple problems, and in these cases the equations are very well behaved. The solving process is actually very logical to the point where you realise that all you are doing is Solving F=ma.
The problem comes down to turbulence, and the fact that the simple Navier-Stokes model do not capture this phenomenon at all. There have been very complicated proposals to the NS equations which take turbulence into account, but these are loosely based on analytical physics and more empirical solutions. Introducing this does not only create a more accurate solution, but employing some numerical trick also make the solution very stable.
Also, there is also the problem of the DE itself. Its not simple to solve, and the numerical methods we usually employ to obtain approximation, are exactly that: approximations.
So if you read the problem statement more carefully, you will realise that there is no straight forward problem that has to be solved. It's like the people didn't know what to set as the problem itself, which has become the problem. To essentially solve the millennium problem, you would need to come up with some form of proof that the NS equations are truly the underlying physics of a fluid (or not). Remember I mentioned the problem of ill-conditioning? Well even if that is true, that does not mean that the NS is BS, and the turbulence modelling tricks can make the solution very stable. However, these tricks are sometimes based on nothing more than: 'it works'. This is not progressive work and that is what the millennium prize tries to address.
So answering the question in terms of your words, we don't know if the solution (real) is smooth. because of which we don't know if using tricks to make our modeled solutions smooth is the correct thing to do to obtain meaningful answers either. And upon finding out whether or not it is, we'd also like to know why? Essentially: solve turbulence, because nobody knows wtf is going on.
A second mistake is that isothermal refers to no loss or gain in TEMPERATURE and not heat.
Sorry about that, I definitely got a bit confused when typing up the script.
I'm considering making a follow-up video as to what was wrong with the video and explain what we are actually solving.
I tried joining your server, but it says that I have been banned or something. Could you see to it?
Discord tag is Napoleon Bonaparte#1729
vcubingx:A cite like this for math, physics, chemistry is not the place to discuss politics including this "b.l.m."!!!
@@roberttelarket4934, why not? His channel, his rules.
@@JivanPal: It may be his channel but it's MY RULE!
@@roberttelarket4934, and thus, your rule is one that no-one is obliged to follow. It's also utterly daft.
Very nice Vivek
Thanks Jens
🙄
Pappa is eager to solve this and win a millennium prize
Can you solve it papa?
Papa Flammy
Because of Your guidance
I know theory of everything Now
" [Universe in a Nutshell] = 42 "
Isothermal refers to a constant temperature process. A process during which no heat escapes is known as adiabatic process.
if there is no heat escape or addition then temp constant only right
@@UnfinishedEngineer lets say a gaz is compressed by a piston, the temperature of the gaz will increase as we have a higher particle agitation due to high pressure, every variation of temperature is concerved as we dont have any heat transfer with anything. That is adiabatic.
If we had colden down the gaz by letting out some of the heat in order to conserve the initial temperature, we would have a constant temperature variation yes but that is because we did a heat tranfer. That is isotherme.
You can have an adiabatic isotherm process.
Iso means ‘equal’ like isometric means ‘equal measurements’.
So isothermal means ‘equal temperature/heat’
Just pointing this out because I literally only found out recently what iso means and it was driving me crazy beforehand trying to remember the meaning of these names.
The exchange occurs slowly for thermal equilibrium in an isothermal process.
Thermodynamics
Kids today that have a natural inclination for maths live in the golden age of learning
Im a med student. Wish i could understand maths easily. Seems much more interesting than human biology. It's very hard for me to grasp these concepts but im not giving up.
@@HexoseMonophosphateCunt Oh man, you must have a lot of free time. I am considering to let go med school to study physics. But ,regardless of what happens, doctors are always welcome. Good luck.
@@everab1209 No man i don't really have lot of free time it's because of covid im stuck at home and have more time, So instead of watching netflix and stuff i prefer learning new things. Thanks though :)
@@HexoseMonophosphateCunt It is good to see people interested in physics despite his main aims. Good luck man.
As a professor of Mathematics this comment is spot on. There is so much information for students at their disposal at any given time.
"In terms of divergance we have no divergance." - Gru
We have to face our karma curve at some point so divergent is something like diversity but we followed the same space so we got equilibrium at some point
@@anilsharma-ev2my I just went way over my head.
Nice one.
I love the navier-stokes equations, I'd definitely watch a continuation of this. Good job man I like your channel very much
Thank you!
@@vcubingx I second this!
Is this something I can argue I need Threadripper for my desktop pc to calculate these? So hard to find any game to actually do demanding calculation, other than synthetic Pi calculation.
This is the best overview of the Navier-Stokes equations that I have seen. The intuitive explanations were very helpful. Thanks!
Thank you Carlos!
@@thealienrobotanthropologist was it really ???
This channel will be having 1M subscriber in 3-4 years .. I got this after solving Navier Stokes equation
Solve again correctly
@@prateekgupta2408 that might lead to chaos!
Some already pointed out mistakes, some key information left out, but overall a nice video. Having tried myself, I know how difficult it is to make videos like these with Manim, so congrats. Also, nice to see more people doing videos on math subjects.
Is NS equation is applicable for laminar flow only or for turbulent as well?
@@brijeshpr6543 It's applicable for any flow, both laminar and turbulent, but it's complicated, there's a lot of practical limitations regarding numerical simulations. Turbulent flows often require a very refined mesh for numerical analysis or some sort of turbulence modelling that are usually not derived from first principles. Computational Fluid Dynamics is a very interesting, but very complicated, area of study.
@@brijeshpr6543 for every type of flow, laminar, turbulent, compressible, incompressible, steady, unsteady. Just the form changes. For example, in this video mass equation is simplified to incompressible flow, and he pointed it out.
ugh millennials and their problems..
lol
Millennials and their million dollars
ruclips.net/video/IFHz-5jsI3Y/видео.html
So when I solve it, will it be navier - stonks?
U should go now lol
That is not the definition of smoothness, smoothness means that it is infinitely differentiable. (Whatever that means) It comes to the study of functions on smooth manifolds, hence smooth functions. The pendulum, for example, I’m not sure that it’s solution has a closed form, but Banach Fixed point theorem assures us that there is a solution!! And it is smooth!!! Now, you can ask then, what would it mean to not be smooth? Well for example the absolute value is not smooth since it is not differentiable at 0. But more than that, experiments on turbulence have shown that turbulence in fluids looks like a fractal!!! And let me tell you, fractals are not smooth in general!! In my opinion turbulence shows us that there is a loss of structure (again, whatever that means).
Thanks for your comment Jose! I was typing up a correction to this at the same time as I saw you're comment. I got confused with a few concepts when typing up the script. I should pay more attention and run it by a few people next time.
Now since you read my comment, I hope you read this one too. Great video, you have a lot of talent and I encourage and celebrate it! So congratulations, and please keep doing videos!
vcubingx I am just a graduate student, but if you can contact experts to check the drafts for your videos, it may help to rise the quality of your work even higher! Again great job and thank you for your excellent effort! You can contact me and I can answer your questions if I can or even better, get you directly with the great professors from my university.
@@josemanuelmedeltorrero7622 Thank you! I'll keep this in mind when I make my next video
Ll
I'd love to see more in-depth coverage of Navie Stokes and fluid mechanics. The video helped me understand so much. thank you
Pretty amazing video graphics! Good work!
Thank you!
These are not the Navier-Stokes equations but rather the initial startup of Hagen-Poiseuille equation. You have forgotten the nonlinear convective acceleration term u⦁∇u on the left hand side, which is what this price is all about in the first place. This term is responsible for turbulence and the white water you’re referring at in the beginning of this video. It should be like this:
ρ(∂u/∂t + u⦁∇u) = ∇p + μ∆u + F
Or with material derivative
ρDu = ∇p + μ∆u + F
Or more commonly
∂u/∂t + u⦁∇u = ∇p/ρ + ν∆u + F/ρ
Where ν = μ/ρ is the kinematic viscosity.
It’s a great video though. Time consuming or not, I would seriously change that, because significantly different equations, more than million dollars to say at least.
I noticed this too. Thank you!
Dude doesn't know what he is doing
What is your insta can we connect ?
I see someone has taken continuum mechanics for fluids in grad school.
Love the color scheme, keep it up with your videos!
Thanks!
Great work I always use RANS (Raynolds average navier Stokes equation) but never had this much clarity of it.
Thank you for this video! I always have wanted some introduction to those equations and now it’s done in a nice and concise way 👍
Glad it was helpful!
Navier-Stokes one of the best ways to scare prospective engineering students.
I managed to flow through that quite smoothly. T.u.
讲得太好了，好详细好生动！感谢老师
Nice animation and clear explanation! Good stuff!
Thanks!
Great stuff. Also, I commend your boldness on tackling fluid dynamics in an accessible way!
Thanks Lucas!
Are you the same Lucas I follow on Twitter.
Similar profile picture
@@nadiyayasmeen3928 Yes, that's me.
I also know manim a lot.. but how do we create and show particles in that vector field?.. can you please tell me ?
Yes! Thank you so much for this video! I’ve been waiting for this for forever!
You're welcome! Thanks for watching!
I will be happy if you make a series about the 7-millennium problems, with this kind of visual representation.💕😍
Great video, clear and deep at once, loved it, thanks for it
Thanks!
Oh my god I’ve been wanting to learn about this for so long
Ah yes. The beautiful Navier-Stokes equations
When they say prove the solutions are smooth, does it mean that the solutions are smooth but we can’t prove it?
As you said we can’t predict weather too many days ahead, so that means the solutions are chaotic but we haven’t proven that either?
Can chaotic solutions be smooth?
From what I understand (and don't quote me, its been a while since I've dealt with fluid dynamics), the problem with the Navier-Stokes equations is the issue of them being ill-conditioned. By that I mean that a small change in the input does not result in a small change in the outcome. This is important since with any system, a small input change should always yield a small output change, otherwise the reliability of the solver is questionable (the results should be reproducible, and near infinitesimal changes should not result in drastically different answers).
Now while the the equations are basically glorified F=ma equations, which means that they are most likely the correct DE that describe the underlying physics, the problem lies in the fact that we simply don't understand or appreciate them enough. Also, remember that the real underlying physics is much more complicated. We can break down the problem to its core where we consider all the fundamental forces of the universe and the quantum effects between each particle in the fluid. But, this is meaningless because we want a meaningful compressed description of the physics, similar to how Newton's laws of gravity are a simpler version of Einstein's.
I've solved the Navier-Stokes equations by hand in undergraduate classes for simple problems, and in these cases the equations are very well behaved. The solving process is actually very logical to the point where you realise that all you are doing is Solving F=ma.
The problem comes down to turbulence, and the fact that the simple Navier-Stokes model do not capture this phenomenon at all. There have been very complicated proposals to the NS equations which take turbulence into account, but these are loosely based on analytical physics and more empirical solutions. Introducing this does not only create a more accurate solution, but employing some numerical trick also make the solution very stable.
Also, there is also the problem of the DE itself. Its not simple to solve, and the numerical methods we usually employ to obtain approximation, are exactly that: approximations.
So if you read the problem statement more carefully, you will realise that there is no straight forward problem that has to be solved. It's like the people didn't know what to set as the problem itself, which has become the problem. To essentially solve the millennium problem, you would need to come up with some form of proof that the NS equations are truly the underlying physics of a fluid (or not). Remember I mentioned the problem of ill-conditioning? Well even if that is true, that does not mean that the NS is BS, and the turbulence modelling tricks can make the solution very stable. However, these tricks are sometimes based on nothing more than: 'it works'. This is not progressive work and that is what the millennium prize tries to address.
So answering the question in terms of your words, we don't know if the solution (real) is smooth. because of which we don't know if using tricks to make our modeled solutions smooth is the correct thing to do to obtain meaningful answers either. And upon finding out whether or not it is, we'd also like to know why? Essentially: solve turbulence, because nobody knows wtf is going on.
I think that last paragraph addresses the question you had about chaos?
Awesome reply Jay! Thanks for this
Jay Raut i understand it very well
Awesome videos bro, hope the channel keeps growing!
Appreciate it!
That was a great video for this topic .Thank you so much for sharing with us .
its like the butterfly effect. a small change in the system adds up over time and makes something we can't predict easily.
Precisely!
I'm pretty sure that's the chaos theory you just described, this on the other hand shows small changes do add up but don't drastically change the outcome. Please correct me if ive misunderstood
The million dollar question is why small changes don't result in drastic outcomes overtime. I think it might have something to do with the correlation between the area said newtonion fluids are operating in.
We can predict streams via geography. Maybe aerospace is harder because of the vairing outside pressure and gravitational changes through a flight
@@slikclips2966 where is the proof that small changes don't change the outcome drastically? i think the more time passes the more change will happen.
I'm actually doing a research paper for the Navier-stokes equation!! Very complex but very fun to read!
I agree, they're really fascinating!
Very nice one! I have derived the whole NSE as well as the Mass & Energy conservation on my channel to actually grasp the concept of where these equations come from a bit better. You did a great job in explaining the main ideas and problems under 10 minutes, props! :D
Thank you! I was a bit worried that I may have squeezed it in a bit too much, but I think it worked out well!
When looking at Navier-Stokes the fundamental properties you are looking at are bulk properties and are impossible to define as a individual atoms. The infinitesimals are assuming a continuous fluid where there are no such things as particles. Think of density in the context of a particle, outside of the arbitrary area that defines that particle the density would be 0 and thus the system wouldn't be continuous.
Rarefied gas dynamics is the feild of fluid mechanics where a gas is treated as a random assortment of molecules. And uses a variety of methods to figure out fluid flow when molecules are so far apart these bulk properties break down.
Thank you 😊 I learn a lot from your channel!
Why does this look so much like 3blue1brown
it uses manim, the python library that 3b1b created and uses
@@stephenhu2000 Just a question: is manim used for the animation or for the math?
@@conanichigawa manim is used for animating and it employs a lot of maths on its own for the animation in the first place
@@AnindyaMahajan Thank you for answering! I was thinking of learning python just for this types of animation.
@@conanichigawa github.com/3b1b/manim have fun!
Pretty onpoint use of Manim. Nice video
Thanks!
2:05 Correct me if I’m wrong but an isothermal proces just means that the temperature remainins constant, not that there is no exchange in heat.
In fact an isothermal process means there is no change in internal energy, which through the 1st law of thermodynamics entails that the work done by the system is equal to the heat gained by the system (I believe that was the correct phrasing of the first law given the change in internal energy is 0).
So if there is work being done at a constant temperature there must be heat gained or lost.
Yep you're right! I corrected myself in the pinned comment
vcubingx sorry didn’t see it.
Nicely explained. So I liked it and shared it. I am already a subscriber.👍❤️
The equations you are showing represent the incompressible Navier Stokes equations, where flow density is assumed constant (Mach < 0.3). This is already a great simplification of the physics and this subset of the equations will not apply to flow over commercial airplanes (Mach > 0.3) and certainly not to rockets (Mach > 1). The full set is comprised of 5 PDE's, conservation of mass (1), conservation of momentum (3), and conservation of energy (1). Solving these equations numerically by marching them in time from an initial flow condition is relatively easy and straightforward, yet it requires significant computing power.
The name of the professor that solved the Navier-Stokes equation is Dr. Gabriel Oyibo
Nice
Just subscribed. Thanks making such detailed informative video.
2:27 We are not describing the behavior of individual molecules of fluid through Navier Stokes equation. In fact, the velocity of individual molecules can be much higher than the flow velocity. Kinetics theory of fluids deals with that topic. In deriving the Navier Stokes equation, we rather treat treat the fluid to be a continuum.
Coolest presentation of the good old N-S Equations. Here , have my upvote .
damn thanks to you I finally understood why div(u)=0 when a fluid is incompressible. Thank you
In a real fluid divergence is not zero because you can probably imagine how if you compress it all into the center, the invisible particles WILL bunch up in the center, meaning that there is more mass entering the center area than leaving it. It all makes sense!
Thanks for explaining the fomular!
I MUST RECREATE ABSOLUTE PERFECT INTELLIGENCE IN THE ALL-SPHERE.
You know the subject is unimaginably hard if there’s no tutorial from our lord and savior the organic chemistry tutor
Fabulous video on this topic. I am learning fluid mechanics this is very helpful
Glad you enjoyed it!
I've been working on this project since quarantine started and have made so much progress, so this video came a little bit later...
but I was more interested in CFD and calculating things through code. Luckily, after soo many hours put into research, learning all this calculus stuff (currently in 10th grade so I had barely any experience with PDEs lmao)
I finally got some C# code working with a Windows Form that allows me to specify the initial velocity, pressures for each cell and can tell me the next frame. Personally a great accomplishment. Something Ill definitely be putting on my college app for my projects during quarantine haha
Thanks for making the video!
Nice job!
hmm - so the goal is to prove that there are solutions to the navier-stokes equations that are non-chaotic? because i've heard a different definition of smooth: not as "opposite of chaotic" but rather as "infinitely often differentiable". or are these two things related?
You’re right - smoothness refers to infinitely differentiable, not sensitivity to initial conditions. In fact, we already know that the NS equation is chaotic and exhibits wildly different solutions for arbitrarily small changes in the initial conditions, so that wouldn’t be a very interesting million dollar problem! To win the million dollars, you need to show existence and uniqueness of a smooth solution, or show that a solution to the NS can breakdown (diverge to infinity).
@@talrefae97 yes, i know that the solutions of NS equations are chaotic in general - but that (by itself) does not necessarily rule out the existence of non-chaotic ones. so, i thought, maybe the quest is to hunt these down, if they exist. like, for example - as happened with the 3-body problem:
ruclips.net/video/et7XvBenEo8/видео.htmlm16s
yes, it's an ODE - but maybe something similar may happen for PDEs as well? anyway, thanks for clarification. or maybe i'm misunderstanding the sense of chaotic here - if it means sensitive to initial conditions - it could very well be that these periodic orbits of the 3-body problem also qualify as chaotic, despite being perfectly periodic. ...i guess that would depend on whether these periodic orbits are attractors ot repellors. edit: ha! here's more about that!
ruclips.net/video/eqSPvyaxMI8/видео.html
so, some of these orbits seem to be indeed attractors (that's the same thing what he means with "stable", right?)
@@MusicEngineeer This is not the same as what needs to be done for Navier-Stokes. With ODEs we have standard existence and uniqueness theorems which tell us any initial condition belonging to an appropriate set leads to solutions that are differentiable in time, at least locally. The problem with PDEs is that the set of initial conditions becomes spaces of functions (as opposed to vectors for ODEs) and so you have to prove that for any initial condition from an appropriate choice of these spaces (typically a Hilbert space) leads to a solution of the Navier-Stokes equation which 1) exists for all time and 2) remains in the space that the initial conditions belong to. It's the second that is the hard part since the definition of the space requires a bound on the spatial derivative, which is currently what cannot be proven.
To get more technical, what can potentially happen is that you have an initial condition that leads to a solution which exists and is bounded for all time, but its derivative blows up at a certain point in time. It doesn't have to blow up as in go to infinity, but blow up in the sense that its norm with respect to some space of functions goes to infinity. That is, the derivative "leaves" the space you started in.
Hence, to prove the millenium problem you need to show one of the following holds for the Navier-Stokes equations: 1) All appropriate initial conditions lead to solutions that are bounded and have bounded spatial derivatives or 2) find an initial condition which leads to a solution for which the derivative blows up.
@@jasonbramburger The official problem states one needs to show that the solutions are smooth (infinitely differentiable) and do not grow at infinity.
The goal is basically to show that for any given initial conditions there is a single set of evolution functions, which also happen to be smooth functions (differentiable)
This related to chaos since numerical methods are about approximating a solution, and a solution needs to exist for this to be the case. And this is not at all trivial. There are plenty of PDE that are proven to not be solvable for arbitrary initial conditions
He never misses.
yessir
Thank you so much for the explanation
Brilliant work my friend!
Thanks a lot!
I hear fluid mechanics, I click like.
this video helps a lot, thank you!!
Source of scaled and shaped flows accumulates heat and tension, so we cannot describe, or solve it, but we can fell it...
We all saw the movie “Gifted” so we are all experts in this equation.
Div(u) = 0 translate the volume conservation. You can talk about mass conservation only if the density is constant in time and space. Which van you have if you consider the fluid is both incompressible and homogeneous, the latter not being specified in the video
Heads off to you bro
Amazing explanation
You are really amazing, go ahead, you gonna be our new 3b1b
Awesome video. Keep the good work
I couldn’t solve the equation yet so no million dollar for me. But Your content just earned you a sub.
when you divide by volume to get density ...you should divide on both sides...
Yes, you spotted one of the big technical mistakes. F cannot be force, it is force density.
brilliant work vivek
My guy, you did an excellent coverage in this very hard topic, but I don't want to be that guy, but here we go. At 2:06 Isothermal is when the temperature stays constant, but Adiabatic is where there is no loss or gain of heat. but CMIIW
3:50 Div u is part of the continuity equation, not the Navier-Stokes - simple Wikipedia would tell you that. Navier-Stokes speaks to momentum conservation.
It's true but the continuity equation is usually included in the pack of Navier-Stokes equations because you need it to close de equation system.
Not the point of the video but in the past four years of engineering school using div , this is the first time someone actually explained what exactly is the divergence operator and what it stands for lol
Would the sextillion dollar equation be some sort of conjunction/unification of the Navier-Stokes & Black-Scholes, along with some other equations?
Brilliant! You should get paid by the ministry of education for that! All faculty teachers should use your videos to teach their students (same for 3blue)
Cheers,
A physics student
Thank you so much! This made my day for sure :)
You're another guy who does work like 3Blue 1 Brown using same elegant animation style
Good video, but I feel compelled to point out that in your explanation of a newtonian fluid is, in a strict sense, untrue although I think you get the right message across. Viscosity is an intrinsic property of the fluid. In other words, the viscosity of ketchup doesn't change regardless of whether it is in motion or motionless. What you really meant to describe was the change in the viscous stresses. Again, this probably doesn't matter for the sake of what you are trying to point out, but is definitely important for someone trying to learn these things in more detail.
And viscosity must not be a scalar, it is a 2nd rank tensor (a matrix). In Newtonian fluids it just happens to be a constant multiple of the identity matrix, so we sometimes think of it as a scalar.
manim!!! thanks for this informative video. im a topology guy so it was a nice peek into pde world
Glad it was helpful! I'd love to cover topology one day
yess maybe you could explain Hodge conjecture using simple geometric analogies about determining all shapes (~= homology classes) of algebraic varieties.
That was fantastic. I wish the video was longer.
Awesome video! Thank you!
Too good
Nice representation of the equation!
Thanks!
So this question occurs to me. is the source or cause or friction in a fluid always the same thing? Just thinking about honey, what are the properties of that substance which cause it to have friction? I know that the friction of honey can be reduced by adding in energy (heat), but I image that there are in fact liquid systems which will not respond to heat that way, or at least I know that some liquids have less friction based on something that doesn't have to do with heat, like motor oil, which heat affects, yes, but at room temperature it is less viscous than honey, so that has to do with something about its molecular make up, not how much heat is in it. (I think) I only ask, because I question the assumption that friction can be thought of as a universal thing, even if the effect is the same, the cause is different, and it feels like that needs to be factored into the equation, or like different liquids would need a unique equation for their properties.
Hey man!
This was a really great explanation.
Thanks!
Brilliant explanation thankyou
it's wonderful! thank you.
That's great... ur video and the equation
Excellent video, thank you !
Glad you liked it!
Nice explanation, thank you
I saw the thumbnail and thought 3blue1brown just uploaded a new video
Great explanation, thanks!
I think that this equation is used to simulate fluids in 3D software like Autodesk’ 3D Max and Blender
Not per se. Numerical solutions are not great (That's kinda the point, they are very chaotic), and they are for incompresible fluid, which is not ideal in simulations. So blender and so on probably use FLIP or other numerical methods, same principle, but very different concepts!
Sorry but one of my pet-peeves for the NS is when people say "We can replace mass with density".
That's NOT what's happening! What you're doing is dividing BOTH sides of the equation by volume, so you're replacing F = m a with F / v = m a / v.
As you say, m / v = density. But what can we say about F / v? Well, F / Area = Pressure. So F / v = Pressure / Length.
Hence, your F / v side features terms such as "Del p" (where p = pressure) and "rho g".
Note that "rho g" is equal to a [ Mass ] / ( [ Length ]^2 [ Time ]^2 ), i.e. the same dimension as Pressure / Length!
The definition of smoothness of solutions given in the video is incorrect.
Exactly. I was hoping that I wasn't the only pearson noticeing it. Chaotic systems like the double pendolum are usually smooth.
What he mention is not the smoothness properties, but the stability under change of the initial condition (different part of well posed properties).
you are a million dollar man, keep up the good work buddy
Good video, first explanation I can understand. Have you used the 3b1b library? It looks very familiar.
Fascinating
Love this video
Oh most excellent video. I see you're using manim. I gotta learn it!!!
Yes! Thank you!
Great video, well done.
Maybe answered elsewhere, but in the intro you say that the NSE can model any fluid...including air, but at 1:52 you say it depends on the fluid being Incompressible. Air is a gas, and near enough under normal conditions, to be an ideal gas - which by definition is NOT incompressible. Its volume always changes directly with its pressure according to PV=nRT, and, if T is allowed to remain constant, the volume changes inversely proportional to its pressure ( V = Const/P). How can both conditions be true? (And no. Air is not "effectively incompressible below Mach 0.3" - when you pump up your tyre, its volume changes with pressure - and nothing is going remotely close to Mach 0.3!)
Excellent video!
Glad you liked it!
At 4:45, the second Navier stokes equation has the term F as the external force term. I’ve watched videos from Numberphile and searched up the equation but I keep finding the external force term being either ‘Fp’ (just take the p as rho, density’) or ‘gp’, etc. could someone please explain to me which is the actual right one? Cuz if you jsut divide F by V, you will get density x acceleration in reality.
Nice presentation. Please may I know how u did the animations and which software tool you used. Thank you very much.