How Physicists FINALLY Solved the Feynman Sprinkler Problem - Explained
A 140 year-old physics problem may have just been solved...Can a sprinkler work and spin in reverse? Comment your answer below as I take a look into this breakthrough research experiment that claims to solve the mystery, once and for all...
Read the paper here:
Centrifugal Flows Drive Reverse Rotation of Feynman’s Sprinkler; Kaizhe Wang, Brennan Sprinkle, Mingxuan Zuo, and Leif Ristroph, Phys. Rev. Lett. 132, 044003 journals.aps.org/prl/abstract...
Merch!
I think Scientists are Rockstars so I made t-shirts to celebrate it
Einstein Rockstar Tee: www.drbenmiles.com/merch/p/ro...
Curie Rockstar Tee: www.drbenmiles.com/merch/p/ro...
Schrodinger Rockstar Tee: www.drbenmiles.com/merch/p/ro...
#breakthrough #physics #science #mystery
Chapters:
00:00 What Is Feynman's Reverse Sprinkler Problem?
0:48 The History Of The The Feynman Sprinkler
3:32 Why Does A Sprinkler Spin?
6:42 Suction Vs Blowing: Airflow & Velocity
8:17 The Experiment
12:37 The Results: Mystery Solved?
13:41 Explanation and Visualising The Results
If you enjoy the channel and want even more physics, tech, and business content, I've just launched new Instagram and Threads pages. Follow on the links below
Insta: / drbenmiles
Threads: threads.net/drbenmiles
Newsletter drbenmiles.substack.com/
A few people have asked so I've added the info below. Some of these are affiliate links. If you make a purchase it doesn't cost you anything extra, but a percentage of the sale will help support this channel and my work to bringing entrepreneurship into science.
My camera : amzn.to/3ed5Xac
My lens: amzn.to/3xIAZyA
My lav: amzn.to/2SeE20Y and amzn.to/3nK33wA
My mic: amzn.to/3gUYYEv
I think if your sprinkler is underwater then your grass is probably wet enough.
Thats why you run it in reverse.
😂
This is why they want to pump the water back out... ;)
"If SOME is GOOD and MORE is BETTER then absolutely TOO MUCH should be just about RIGHT." Had some tee shirts made up years ago memorializing a meeting where a senior VP went around basically chanting the first two-thirds of this quote in his presentation. When I added the last third during a lull in the chanting, the VP just stared at me with a stunned look. The President took a liking to me right then and there and made sure I was included in more meetings, which were occasionally not fun.
Yeah. And besides needing a reverse sprinkler you'd have to invest in a seaweedwacker.
Imagine being so smart that a Problem gets Your name because you could NOT solve it.
Sounds more like an ego problem than anything else
I think Mr. Sprinkle got involved in case it gets to be known as the Sprinkle Sprinkler.
I don't see the ego problem when Feynman didn't name it himself.
Sure thing, Einstein
Most mathematical problems that are waiting for a solution are named after smart mathematicians who could not find a solution. Q.E.D. Once the problem is solved, it does not take on the name of the successful first solver.
When I was a physics grad student in the 80s I disagreed with a professor about an E&M problem - the prof was a real *sshole about it and I was sure I was right. I phoned up Feynman at his home (he was in the directory!) and asked him his opinion. He told me I was right (this story ended up doing the rounds at UCI) and he asked me the sprinkler problem. I gave a few different answers that I said were naïve answers (which are covered in your video!), and that I was unsure. He told me to call him back when I had my answer. Overall we had a 45 minute conversation - I felt very honored. I became disappointed in myself as I never got a fully convincing answer so never called him back, and he died in 1988. I felt like I had failed the great man - until I saw your video today!!!!
that is awesome, having been able to ask feynman about your problem :D
I was a chem/physics student at UC Irvine in the 80s. Any chance you could hint at the prof's name? (Edited to clarify university).
Rest in Peace
Am I the first person to notice that the description of Feynman's experiment is wrong? Actually, he tried to pump air into the top of the carboy to push the water backwards through the tubing; he didn't suck the water out of the tube. Eventually the pressure blew the carboy apart. See "Surely You're Joking, Mr. Feynman" at the end of Part 2: The Princeton Years.
Makes a lot more sense, I'm sitting here wondering how on earth he broke the tank by just sucking in water
It took 140 years to put a sprinkler underwater.
lol
precisely.
Well the optics might have been available to scientists 140 years ago but the lasers are a more recent invention. And the computer required to crunch the data for PIV even more so.
@@salsamancer none of which is needed to put a sprinkler underwater
"Feynman was keenly aware of his own abilities and almost entirely unburdened with modesty" - the sentence where I clicked Subscribe.
For me it was this one: "... and this year's entry for nominative determinism, Brennan Sprinkle."
Yep, that was firmly in second place for me too!
and a huge sexist at the same time
@@oxiosophy let me guess, you don't believe that Feynman was a great person academically and otherwise, right?
How does modesty burden one, except for the burden on the ego?
"almost entirely unburdened by modesty." That is the greatest description of Feynman! He wasn't so much arrogant as bereft of any desire to *not* be arrogant.
I think you're not quite right. He really really wished to avoid being given any credit for one skill on the basis of something that had nothing to do with it. He didn't put his name Feynman on the drawings and portraits he was good at.
Me, skipping randomly on work through the video: 7:37 _"[...] Interestingly here due to our slightly imprecise use of language when we describe sucking and blowing [...]"_ With a humor stuck stil in puberty this line without context humours me a little.
They also called the simulation PIV...
Why didn't they repeat the experiment with internal tubes pointing upwards to cancel the vortex?
For the same reason it took them so long to.....just freaking build an apparatus and test it..... Because physicists aren't as smart as engineers ;) Id argue if this is the effect at play then they obviously could manipulate the tube runs to reverse the reversed reversal of the reversed flow....ya know, but backwards.
exactly my point - you said it clearer.
@@TankR Yeah they could've just used the local swimming pool at the deep end. Mythbusters would've.
I just wrote that above. You beat me to it lol. I have a couple of variations in my comment. One was to use pressure instead of suction.
yes, it shouldn't matter hey.@@ThePaulv12
I'm giving you a thumbs up for excellent audio quality, no over powering music and clear responses. Great work here.
* overpowering
I don’t think he needs an explanation for every like
That meniscus bearing is cool. I wonder where this idea came from? Can this be used to create frictionless bearings for more practical applications?
Magnetic bearing: No.
Fluid bearings like this are common in industry, both water, oil and the classic air bearing.
@@otm646 FLUID bearings are common, but I don't think this concentric-MENISCUS bearing is -- did you actually look closely at how it works? I don't think it would provide enough radial stiffness for any uses except sensitive instrumentation.
i had a feeling the fan topic was gonna be brought up and lo and behold, 6:42 comes up
I wish they would have redesigned the test so the arms of the sprinkler don't have that central cavity for the vortexes to form. They could have brought the two tubes together in an upside down Y with the leg of the inverted Y pointing straight up in the center -- that should eliminate those vortexes that were contributing rotational forces.
Yeah, after watching the video, I still have the feeling that this massivly depends on thd design of the sprinkler. Of you'd add an infinite ammount of arms, you'd end up with something simmelar to a tesla turbine, which would possibly spin in the other direction. And a different hub layout might also form the center vortecies in a different way.
my thought exactly. they could even bend them parallel before joining along their sides to preserve laminar flow. to the point that even there the cumulative outer track of water would still move faster and might still cause slight asymmetries: then there is probably a way to angle the inlet jets entering the central chamber to compensate for the lopsided velocities. just angle them until all 4 vortices are equal. there's probably a million variations you could build, but I'm betting per design, there is a small adjustment which preserves the behavior with water flowing out, but which is balanced when water flows in. its not about principle, its about which of like 7 minute balancing acts your current design happens to be failing the most, and that is the latent rotation being seen.
@@clockworkvanhellsing372directional baffles could align all the vortices, I would think. Such a setup should eliminate now dampened speed, and making it more relative omnidirectionally.
The tip of the nozzle has a lower pressure than the surrounding water will pull the nozzles forward
Excellent thought ALSO the fact that the outer curve away from the center has more surface area against which the incoming fluid would press against would seem to be relevant too (I’m not sure why it would matter vs fluid already in the tube though to be clear…) - even altering the laminar vs turbulent friction with varying materials there would impact things etc!?
This seems more a function of the specific design of the sprinkler internals. If the pipes were angled to have the vortices' sizes inverted it could be made to rotate in the other direction.
If there was a suitably shaped baffle inside the sprinkler head, it could be therefore made to rotate slowly in any direction, or not at all, it seems. Normal sprinkler rotation, including overcoming friction, is surely mostly rocket science - reaction to the mass of the water sprayed in the opposite direction of rotation of the sprinkler head.
They should have directed the internal jets upwards so all the incoming water streams are flowing in parallel before they are allowed to interact to avoid "spooky actions in a (hidden) vortex". IMO this experiment has not effectively addressed the original question.
Yea, I have no idea why they had two tubes in this experiment. I assume three would cause similar vortices but how about just one tube so there wouldn't even be a need for a chamber like that? Could even have just round corners so there shouldn't be any noticeable vortices at any point.
Yeah, that's what I was thinking. I want to see what happens if they design a system to nullify these internal forces, and focus only on the water in the arms of the sprinkler.
@@lisabenden If you nullify the internal forces then you aren't actually testing the problem :) The internal forces are a result of the bends in the pipe. If you nullify them, then you will have to by definition design your "reverse" sprinkler, to impart forces onto the sprinkler system to counteract the "natural" designs rotation. By having pipes that bend, you end up with water flow that is not "centrered" in the pipe. This "non uniform flow" is the effect that causes the rotation.
A while ago I saw a KZhead video that immediately came to mind. My first thought was also that pressure is equally everywhere in every direction, by the way. But the video was about a simple vertical (PVC) pipe connected to a vacuum cleaner. It was mounted to the side of a table, but not actually fixated in place. When the vacuum cleaner turns on, the pipe moves up a bit. Conclusion was that the air that is right next to the pipe gets sucked in with a sling-shot motion and the centrifugal force that came with it, pulls the pipe up. It also heavily depends on the shape of the rim: a well rounded edge pulls less.
These are the types of videos that make me glad to study physics in college. I guessed right in the first part, surmised the opposite in the second part, and I was happy with the result in the third part. Always adapting to new information and ideas.
TLDR: The 100 year old answer of "depends on what engineering choices you pick to have the most effect" is the right one and nothing was actually discovered beyond why small house vacuums often have the intake opening on the side which was also known for quite a while.
Ye i thought the same. So if the sprinkler doesnt have cnc quality openings but instead a janky mold of some sort the answer would be completely different? How would the scenario play out if you used turbulent flow?
@@D3nn1sor if you had more than two intakes at various angles.
They basically engineered a sprinkler to get a result. That particular sprinkler design didn't exist until they made it. Its result is rather inconsequential. As the question was concise and the parameters were quite clear, the experiment used methods that eliminated mechanical friction, which exists in all functioning sprinklers. The mechanical friction was not a variable that needed elimination. The question was not "what would happen to a specially designed sprinkler submerged underwater if it sucked in water." It's a nice little experiment, but I don't think it actually answered the question. In fact, the design probably fails miserably at being an actual sprinkler.
No, the question is what is the result of the forces in that system. Mechanical friction and unaligned tubes would obscure the actual results, while this set up is what an "ideal" sprinkler would act like. This way, we discovered what actually had a predominant effect on the direction of rotation, which is the flow in the internal parts of the sprinkler, more than the liquid actually being sucked in or hitting that wall in the first bend in the tube
I would've loved it if they did other designs too to see how the angles and types of flow contribute to those vortexes, but this is still an interesting result
Are you telling me that not _one_ person decided to bend the tubes upward toward the pump-rather than just ending them at cavity where they point at each other-in order to basically remove the vortices entirely? It's like the question hasn't been answered at all, at this point.
Hey English isn't my first language and I didn't understand your suggestion. Could you draw it and send a link?
What are you yapping about
Its simple, the video author explained about how the submerged sprinkler sucks in water and the individual legs of the sucking tubes are ending inside in a mutual opposite alignment (which is the reason for the submerged sprinler rotating backwards) But inorder to truly find out the sppinning effect by avoinding this new disturbance, both the sprinkler tubes can be bend 90 degrees and be taken entirely out from the water so that the problem of momentum interaction of water molecules inside the submerged sprinkler head will not arise. The true motive of the experiment can be served justice. Now did you get the idea ?@@marvin.marciano
It wouldn't necessarily remove the vortices, just change their orientation. Any asymmetry means they could still provide a net force. I would rather see a design with just one nozzle (and a counterweight for balance) with the pipe having, as far as practical, a constant diameter from pump to nozzle.
@@chicklucas6682Are you genuinely unable to visualise what the OP describes?
I think it is possible to analyze the problem in a simpler way by breaking it down. 1/ pumping fluid quickly inside a simple tube with an entry and exit generates strong pushing thrust at the exit, and weak pulling thrust at the entry. this can probably be measured independently using load cells. the thrust can be converted into movement/rotation or a stationary force/torque, it doesnt matter. this is highlighted by jet ski having the jet exit direction controling the thrust, while the intake is directed forward and downward (not straight forward) and doesnt change direction for forward or reverse operation 2/ if you now have several exits, and several entries, the overall thrust will be approximately the sum of the exit thrusts 3/ if exit thrusts cancel each others approximately, then the intake thrusts can become prevalent 4/ if exits streams point at each or at fixed objects other weird turbulence and vortices will happen and create additional secondary effects way more complicated to study and probably cant be predicted without numeric simulation and understood through experimentation 5/ even it the main exit thrusts cancel each other, those secondary effect could still outweight intake thrusts. THIS IS probably the ONLY CONCLUSIION of this experiment? 6/ the rotating part of a sprinkler should be analyzed like a freely rotating system with entries and exits for fluid to be pumped through 7/ the traditional sprinkler has several exits which combined generate a clear torque, stronger than any effec onthe sucking side, the intakes don't matter 8/ the generic sucking sprinkler achieved using any sprinkler, with reversed pumping action, is designed wihout any attention to the blowing side , and because of this, has undetermined behavior 8/ the sprinkler shown in this experiment is seemingly designed to cancel the effects of the blowing side to show the effect of the sucking side (by using symetrical exits, pointing at the center, but failed to do so because asymetrical flows and resulting asymetrical vortices
I remember seeing a model of the inverse sprinkler years ago with air being sucked in. The result was that it was sensitive to disturbances and it was possible to get it going in both directions. It wasn't going that far to reduce the disturbances though.
to the people saying they only got this answer because of the way they designed their sprinkler: they also did all of the calculations and math derivations so you can now predict the movement of many sprinkler designs, not just the one they actually built.
Yea that's the most important part of this study IMO. The results were obvious and it's embarrassing this wasn't "solved" earlier as vacuums knew and solved this internal vortex issue several decades ago, thus I already knew the results. I figured this video was going over something that was solved in the 1990s or something but it's pretty sad looking at that date..
Ooh that's neat
I think it's because the geometry of the plenum wasn't specified, and the effect would disappear depending on the plenum's geometry. This seems more like experimental error unless the problem specifically states that the sprinkler has to have this specific plenum geometry. I assumed the sprinkler wouldn't move, but i also assumed the experiment would provide a suction via a 2:1 header with decent flow characteristics rather than dumping asymmetrical flow into an internal volume. Of course it would spin in revere in that case, but arbitrary changes to the internal volume can give any result. Care was taken to isolate the system from pump vibration, meniuscus bearing for lower friction, all this is pretty obvious and what i'd assume would be the setup, but i also would assume that the experiment would account for the internal geometry by using a low turbulence Y connection to the suction. As the problem was being described, i already thought of a siphon and meniscus bearing, as well as a fairly laminar internal structure. The experiment is specifically designed to give this result, and it can give the opposite result if the internal geometry contained baffles, guide fins, a rounded feature in which an axle/pivot bolt runs, or any number of possible configurations. When i design hovercraft hulls i use all kinds of tricks like this to negate lift motor torque by adjusting the plenum geometry.
Haha, I just read your comment after basically posting the same exact thing.
I wonder, if they were able to perfectly match everything to minimize design biases influencing fluid dynamics, would we see outside forces acting to create a spin, such as Coriolis and gravitational effects from mass concentrations.
I'd like to see the arms be offset from the axis of rotation to create an internal vortex that is opposite the observed head rotation direction. Make the arms reversible as well so they can create the forward internal vortex and see if there is a difference in speed. To me the rotation is obviously attributed to different pressure at the suction face and reverse side of each arm. That is where the external system interacts most strongly with the internal system.
In this case, though, isn't the force ultimately from the different total curvature of the inside and outside paths of the pipes? I know that the differential vortices are apparent at the center, but it seemed to me that the force arises as an imbalance in the suction experienced across the inside walls of the pipes. (The different speeds and sizes of the central vortices are therefor a result and not a cause.) It's surely true that you could overcome that by introducing other geometric facts at other locations. However, it would seem to remain true, that in a truly symmetric system, that the result obtained here should remain. Do you not think it so?
@@fnamelname9077In this specific internal geometry a rotation is imparted. The question isn't asked in terms of the effects of the internal geometry, but in terms of what effect the suction has in terms of imparting rotation. In this configuration the signal to noise ratio doesn't allow a valid result in terms of suction. If the internal geometry isn't fixed as a constant and can be anything that causes suction through the arms, i can easily design a range of internal geometry configurations to impart any desired rotation. And as for the differential in the pipe, it becomes a larger effect only because there are colliding differential flows. This will not hold true if there is only one tube, or three, or seven, or if the tubes were flush to the inside, or if those tubes were tapered, or angled, or mitered at the ends or, or and on and on. Unless the internal geometry of all sprinklers are an ansi standard, then all sprinklers will act differently, and the answer only applies t this specific case. The answer is that they answered the wrong question and call it solved because noise caused the result. This might just bother me enough to run this experiment myself with a setup capable of giving a result, it's not like it's hard to 3d print some tubes and siphon some water. And in the case that the confounding factor of internal vortices is accounted for, there's a whole other can of worms with the nozzle shape. Is the sprinkler arm tip just a cut off section of pipe? Does it have an internal taper? Beveled outer edges inducing Coanda effect? Something else like decorative plastic flowers that the water shoots out from? Everyone in Feynman's class disagreed because they all have different brand sprinklers a home lol.
Thank you for this great insight in how to analyze and tes problems, great stuff.
Since there are left and right openings causing four vortices, two big ones are form by collision of two inward jets. two inward jets are formed from left and right pipes (2 direction). The question to ask are 1. what happens if it is 3 direction or 4 direction sprinkles for these experiment? Doesn't it stops? 2. does the vortices being indirectly related to the mirrored direction of the bend pipes? If so can't we create a specific bend pipe that stops spinning 2 direction?
If the three pipes are still concentric, then you'll get three jets meeting at offset angles, which probably will still cause spin. That said, minor changes in the geometry of any part of the system could cause significant change in rotation speed, in its direction and even in whether it happens at all. You can almost certainly get a three-tube model to spin in reverse, in the same direction as when sprinkling or even to not spin, depending on how you build it.
"entry for nominative determinism" slayed me sir. bravo
I remember hearing about a problem that early jet aircraft had, especially those with intakes in the nose. It was called "inlet lift". At high angles of attack the air entering the inlet had to turn a corner, and this created a nose up force. This made stall recovery tough because adding power, i.e. afterburner, increased the effect.
stall is a lack of lift. and you're saying now that increasing the total net lift is making stall worse.. are you daft? not to mention that same effect is happening at the bottom of the engine aswel in the opposite direction. and thus cancels itself out. you must be on drugs or something.
That answer is actually really fascinating, and it just goes to show you that it's not always outside, but what's inside that really counts
My initial conclusion when hearing the problem was that it wouldn’t spin for the same intuitive reasons that explained why the force from sucking in fluid is much much weaker than expelling fluid. Now I also made an assumption that those tubes that went into the sprinkler housing, didn’t just terminate immediately into an empty cavity where vortices can form. I assumed the tubes would bend downwards. If the tubes did bend downwards once inside the housing, would the sprinkler rotate at all in this case?
8:35 of course, Brennan _had_ to work this problem. fascinating 🖖
There's a long history in the UK of people who's namesake became their job. My metalworking teacher was called Mr. Bolt.
For some reason, YT wanted my “feedback” on this comment.
Knowing him, 90% of his motivation for this was the joke.
It was his calling
Not so convinced.. since you talked about the opposite effects of sucking and inertial forces in the pipe corners, Reynolds should be an important factor. The pressure gradients involved in sucking are influenced by viscosity, while the force imparted on the tube due to the fluid changing direction are not. I expect it would spin in the normal way at sufficiently high Re.
Yeah that and changing the dynamics of the center fluid removal would seem highly relevant etc
I'm satisfied with the above explanation neither. Imagine sprinkler mouth sucking a thick jelly, so it will cut itself into a jelly. And water may behave like such a thin jelly in this regard.
Kind of a random addition to the experiment to allow the fluids to collide inside the sprinkler. This should not be part of the experiment, and mitigated with the pipes being fed / sucked by separate tubes. Or them being bend upwards before joining. The whole experiment lacks a certain clarity of its definition.
@@vast634 Yes, the turbulence effects inside of sprinkler should be eliminated by experimental arrangement in similar way, like the described experiment already does outside of it.
@@vast634 I agree, the internals are irrelevant to the problem. At the very least the internals should have been isolated to be nonconsequential. Otherwise too many variables.
This was real fun to watch. Thank you a lot for this video.
nicely done and brilliantly scripted, illustrated, and produced ty for posting!
It would be interesting to have the tubes inside the central cavity turn so that they are pointing vertically (up or down) compared with the axis of rotation.
This was already solved a decade ago at Harvard. They knew about the votices inside the hub and that the rotation depends on the geometry. Wang's contribution is more subtle and the new "solution" is pop science sensationalism
The Harvard study did not include the internal geometry and how this can change the rotational direction
@@Ghredle You didn't read the paper
Thank you
@@shanent5793 no i did not …just had access to one drawing which shows the water intake… my assumption was based on incomplete knowledge,
Seeing as the real solution (aka depends on what you engineer the sprinker for/emphasis on what forces) was already present during RPFs lecture, it was solved more than just a decade ago and not at Harvard.
This was quite interesting- in the end I found the answer a little confusing, but it’s nice to know it does spin reverse to when water is flowing the other way. Have you ever studied the “chain fountain” problem? Similarly related to momentum imparted, we see chain will rise from a pile as chain is dropped over the edge of a container. Where does the upward momentum come from? I’d appreciate your discussion if this.
That was an eye opening demonstration of forces that I never thought about.
Did Mach's theoretical sprinkler have the attitude of the internal ends of the tubes defined? Thanks to all who work on this problem, it has been spinning around my brain for decades now, since I read Feynman's book.
What a fascinating result. Fluid dynamics - elegantly simple rules that often defy expert intuition.
Alright, I'm convinced that these vortices impart their momentum on the central body, however I'm not quite convinced yet that the negative pressure at the point of ingress is completely overruled by the positive momentum imparted on these tubes by their internal flow's directional change. How about redoing this setup with a longer straight closed-off tube with a 90deg smaller soldered-on jet, and increase the flow to the turbulent realm? That should mix up the flows enough in the core that eddies should be randomized and cancelled, and due to the lack of bend, it'll only be about the effects of the negative pressure at the jet start, and positive at the reaction wall.
Thank you. “Experimental design” questions answered that occurred to me as you were presenting the facts, possible solutions and attempted proofs. Very clearly demonstrated and well explained for a visual, life long learner.
Next experiment: stop those vortexes from forming within the center of the hub
That's what I thought.
So, if you added a 90 degree bend pointing up to the suction area then all rotation should stop. Right?
And you could change the direction of rotation by changing the angle the pipes enter the central chamber, right?
And the direction of rotation actually wouldnt be affected by the external angle of the pipes, assuming the vortices still formed in the same manner, right?
8:00 Wait a sec, did you use battered side down footage??? Awesome!!!
Fascinating! I spotted the first two forces, but (fluid dynamics being largely over my head) didn't know whether the force due to the pressure difference and the force due to water going around the bend would inherently cancel out. Hadn't even considered the behavior of water inside the hub.
7:55 "Timmy, close the window" "Oh, sorry dad."
When water is spit out it all goes one direction (due to its momentum inside pipe) but when sucked in, it comes in from almost every direction (except for the direction of the pipe) since its initial momentum is close to zero. This is why “put-put” boats work.
But they need to make it unnecessary complicated
15:40 this sounds good, but it seems to me that this also means that it should be possible to design an internal geometry for the sprinkler that cancels out the effect, or even reverses it. It should be possible to experimentally confirm that this is really the main effect. It also seems to me that it ought to be interesting to measure the power and energy involved.
It doesn't feel like this is really answering the question. I do not think the original hypothetical was supposed to consider the effect of the internal cavity of the sprinkler. That seems like the bearing resistance issue the experiment was trying to solve for. We establish initially that the normal sprinkler rotates because of the force of water going through the tubes, without considering what happens when the water first comes into the central cavity. So, to ask what happens when water is sucked out, it doesn't seem like we should be looking at what happens when the water enters that central cavity. What would happen if the water was sucked all the way out of the sprinkler system (for example, all the way to the side basin) rather than into the central cavity where the flows press into each other? Does that make a difference.
also, there is no "sucking" only pressure differentials. meaning fluids always get pushed, never pulled.
Well, it's a straightforward problem in the case of something like a bow thruster, which is a "ducted fan": a symmetrical arrangement of a propellor in the middle of a duct open at both ends. It's not hard to argue that most of the force transfer here is at the surface of the propellor itself, which in turn is transferred via the mounting frame to the vessel. In an open environment, very little force can be said to result from the thruster developing higher ambient pressure on one side of the vessel relative to the other. But that "very little" difference is still NONZERO and, significantly, it has the SAME SIGN as that of the blade thrust. The Feynman sprinkler, for obvious reasons, develops perhaps HALF of that pressure difference in the best case. We might say that the entire ambient environment is at a common pressure, but in the area close to the vent, the pressure is lower. If you put your finger over the vent, you can easily feel it being drawn toward the vent. That's a rough measure of the available motive force in this negative pressure scenario. The fluid in that region has mass and therefore resists being accelerated. The various resulting force vectors in the neighborhood cancel except for the component along the axis of the vent. In short, this force may be modest but it is NONZERO, and it has the SAME SIGN as the flow through the duct, which is inwards in the case of a Feynman sprinkler. A broadly conical vent will tend to contain this negative pressure and direct its force more in line with the vent axis. It will still be more diffuse, therefore less directed and effectively weaker, relative to what is possible with a positive pressure through the vent. But if you imagine making the vent into a diffuser, you can see how easily the positive pressure scenario can be weakened as well, until the two scenarios become quite closely comparable.
How does this apply in a situation where you suck on a straw? You are creating a pressure differential between your mouth and the water, at which point the water travels up the straw to enter your mouth thus balancing the differential. I would consider that a pull.
Unless you're talking ferrofluid and magnets.
Ohh wait is it becasue the pressure of the world outside the straw is now greater than the pressure in your mouth and it pushes the water up the straw?
@@brianthibodeau2960 yeah, when you arent sucking the air pressure inside the straw and outside are the same. once you start sucking, theres less air pressing down on the liquid inside the straw vs outside, so the air outside is able to push the liquid up the straw to try and equalize the pressure. if you had a straw going all the way to space (so just a tall straw with a vacuum inside it) it would only be able to push the liquid up a certain distance before the weight of the water in the straw is too much for the atmosphere to keep lifting. so you could put a tube from the ocean to space and it wouldnt drain the ocean
6:40 - yes. Sucking and blowing can be the same thing. However. Context is really important
A vacuum cleaner can be made to suck or blow, however the suction is very local and directed into the head, but blowing is always at a distance, and the effects are much more random, which is why I object to council road and park maintenance operatives using fossil-fuel driven leaf blowers to scatter the leaves in a general direction, before being picked up by other means. If they had vacuum cleaners, the leaves would be sucked into receptacles on site or by hoses connected directly to their vehicle's leaf collector directly.
12-year-old me: "he he he"
One of the issues I've always had with experiments around Feynman's sprinkler is the description at 4:45 of how the water particles impart a radial and tangential force via collision with the inside of the tube to describe how momentum is passed from the water to the sprinkler causing it to spin. If the tube had a square corner, at 90 degrees, the sprinkler would still spin. So how does the side wall collision theory explain or account for that?
1. A particle can not turn 90 degrees instantly, that would require infinite acceleration 2. I think that some water in the outside corner will act as curve for the majority of water to go over
That was actually remarkably informative and entertaining. Thank you.
I don't know who pointed the editor to that Simpsons clip, but you need to give them a big fat raise.
And fire the dude who did the volume levels for that part
It's obvious that it wouldn't spin if the liquid is drawn uniformally from the system (which can be achieved through inner arrangement of the sprinkler) because there is no net change in the angular momentum of the water. Basically the way it spins depends on how the water is leaving the system, not how it enters. Same as with regular sprinkler. Edit: The answer given in the video is only correct if you want to know what forces do sprinkler arms contribute and ignore everything else. Which is not quite the same as the original question
14:40. This shows that there can be a torque depending on how the water enters the central drum. This means if you modify the design of the tubes entering the drum, you can make it spin *either* direction, depending on how much angular momentum is acquired by the water exiting the drain. This should probably be viewed as a flaw in the experiment. If you design the drum specifically to prevent the water from acquiring any angular momentum at the drain, it will not spin. As an example, turn the tubes in the drum straight downward so that water must exit without angular momentum.
Was REALLY looking forward to you going over the "friendly fluid dynamic equations" in Wang's paper.🤭
That sprinkler should be in a museum somewhere with an explanation of the design constraints and how this specific design solves them. That is a thing of pure art
Commenting at 0:53, before watching your report. Many years ago, in the early days of the worldwide web, I was curious about this. I searched online and found groups that reproduced the experiment and even had video! These were done in various ways, e.g. with gas instead of water, and other fluids. Also, the sprinkler arm shape varied. The empirical results were all over the place! Some went one way, some the other, some had no discernible movement. I also found a draft of a paper analyzing the physics, and with simplifying assumptions, concludes that in the steady state there should be no motion. I emailed them with the experiments I had found, and later found that not only had they updated the paper but I got thanked in the list of credits at the end! So, with the forces in opposite directions cancelling, these "simplifying assumptions" will be all that's left. Due to viscosity and friction and the arm shape and whatnot, tiny imperfections cause the cancellation to be less than perfect. Whether the real-world sprinkler moves one way or the other depends on the precise design, the fluid viscosity, friction, resident time inside the arm, and who knows what else. So, if this experiment claims to be the first/only people to try it, they are _so_ wrong. If it claims to be the first detailed analysis of the physics, they are wrong by a few decades. So I wonder what is meant by "breakthrough" and why it's "finally" solved when it was solved more than twenty years ago IIRC. Is this all not on the Wikipedia page, perhaps, so nobody else knows about it? I wonder if ChatGPT knows about it? ...
Yes, ChatGPT remarks, "... This is because the system's behavior when water is sucked in can be counterintuitive and is influenced by various factors, such as the design of the sprinkler, the viscosity of the fluid, and the specifics of the fluid flow (laminar vs. turbulent)." Asking about specific results, it notes, "...small differences in experimental setup can significantly affect the outcome, and theoretical analyses often rely on idealized assumptions that may not fully capture the complexities of real fluid dynamics." and, "...but the direction and magnitude of this motion often depended on specific details like the shape of the sprinkler arms, the presence of nozzles, and the rate of fluid flow."
ChatGPT was trained on lots of content from the internet, so it can maybe be used to compile the general gist of what people on the internet are saying about the topic. Just a little reminder ChatGPT is a language model, not a general intelligence.
Yep, the AI is still far from obtaining it's"I".
@@HenrikMyrhaug the notion that GPT just "compiles the general gist" is straight up wrong. Neural nets can generalize, systematize and extrapolate information. They can deduce something entirely new and can develop task-specific emergent properties. GPT isn't a reliable source because it doesn't do what we expect it to do, not because it doesn't do anything remarkable
@@HenrikMyrhaug This is something of an arbitrary distinction given that it can solve very nearly entirely arbitrary problems, as long as they can be phrased in a textual manner. _So many_ papers are "we presented x problem to a big language model and it turns out it can solve it pretty well."
Thanks for the video! I found the paper too dense for a leisurely read, but this was perfect for my curiosity.
Thank you for this. Amazing visualizations and explanations. I love having my brain scratched.
Yeah, but why would you want to suck anything with a sprinkler...? 🤔😂
It has greater implications to physics and the universe...
Because it is possible ?
"How Physicists FINALLY Solved the Feynman Sprinkler Problem" - I didn't even know it existed.... lol... 😂
One of his books has entertaining story around that--as a student he first convinced his professor that it must spin in one direction then later equally convinced him that it would spin in the opposite direction.
The simplest sprinklers are composed of hoses with holes punctured at regular intervals. Great presentation and delivery of the material. Only 4 minutes in but I appreciate the thought and craftsmanship that has been invested in communicating this problem.
What a banger video! Excellent summary of their paper, there's so much to this
Living in 2024 is phenomenal, in terms of how deeply informative content just casually drops when you may most conveniently want/need it. Just a couple days back I was contemplating the problem in leisure, and after having settled on my answer, looking around the internet for confirmation. To my surprise, there was a paper...lo and behold...from 2024 itself! And to my even greater surprise just now, a KZhead video on the topic tips its hat in my path. Internet was always known to be powerful, but something wonderfully invigorating is making it powerful still!
Maybe there is no sprinkler??
The paper proved there's at least one, as one of the authors 😄
I think the mass difference exiting or entering tube is a crucial reason for the rotation of a sprinkler other that the obvious other reasons and that changes the outcome of suction. It's just amazing they didn't use something else to compare the problem to
Wouldnt the potential direction of rotation depend on the design of the sprinkler and angle at which the arms come into the center of the sprinkler as welk as the internal design of the sprinkler? Another thing that may have an effect, though I'm not sure how or what the effect would be (or whybit would necessarily effect it) would be to set up the sprinkler like a normal sprinkler with the water coming into it from below as if you had set the sprinkler on the bottom of a pool or in a lake. The feed line could still be used as a siphon for suction it would just need to be ran across the bottom of the tank and then up the wall of the tank instead of straight up and out from the center of the sprinkler.
I would expect it to spin the other way , but very slowly because if you've ever tried to use a box fan to "suck air" you know that the air isn't sucked out in a clean column . So suction is NOT the inverse of blowing .
cmon, its somewhat like the inverse of blowing, precisely why i also expect it to go slow the other way
12:59 if you don't want a 13 minute history of sprinklers.
This experiment raises another question in me. If you were to extend the arm pipes inside the hub 90° upwards, would the sprinkler not rotate and instead move deeper into the water? The suction tube that removes the water from the sprinkler is mechanically decoupled from the hub so it should not cause any back-pressure and with the feynman sprinkler effectively being an inside-out regular sprinkler, I would expect my modified setup to simply sink.
this seems like an answer to this one specific case. if there were three arms, would the nature of the vortex lobes in the central chamber be different? and what's the behavior in the limit as you add more and more arms? is the acceleration constant or does it get faster and faster or does symmetry or turbulence in the internal chamber's vortices produce some other effect?
So... what would happen if you angled the inflow pipes within the center of the sprinkler? Since it's the quasi-lamellar flow is what turns the arms during conventional operation, would not angling the inflow lines at the opposite end impart the same rotational force when flow is reversed?
I don't have the math to back it up, but it would seem that there would also be an effect from higher pressure on the opposite side of the pipe as the intake as well as friction on the sides of the pipe as the fluid that's in contact with it runs toward the end of the nozzle before undergoing a 180 degree change in momentum which in itself would have to impart force somewhere.
the quad vortex in the center could also drive torque to the cylinder because the arc length of the fluid swirling in the larger vortexes drags the inner wall of the cylinder around and overpowers the smaller vortex. they should add some baffling to the system to avoid these collisions and re-test. a 90 degree bend from the inlet of each pipe in the center pointing vertical would isolate and converge the flows together better and isolate the variable out of the system.
Perhaps entrainment of surrounding fluid pulls it from the side opposite the holes due to vortex generation around the round tube? The closest water to fill the low pressure void would be touching the perimeter of the opening, so I imagine it like hydrofoils sucking air down along the vertical support?
Mesmerising and stimulating : Thank you for sharing
It seems like you could make it spin whichever direction you wanted by changing the direction of the inlets into the central chamber to something other than oppositional. If they had built the central chamber so the inlet pipes pointed up or down instead of oppositional or left or right you'd achieve a different result by modifying the way vortices form or don't form. They did all this work to basically prove nothing because the design of the system simply shifts the "blowing" effect from external to internal.
I was taught that there are 5 possible methods to any kinematics problem. This one is readily solved with "conservation of angular momentum". Angular momentum is imparted to the water drawn into the sprinkler. The key question is where is this angular momentum shed. If it's also shed within the sprinkler then it won't spin (except initially when turned on). If it's shed outside the sprinkler then it will spin in reverse, as shown. Doesn't seem super hard. Of course, without "conservation of angular momentum", it becomes one of the hardest problems imaginable.
That was a brilliant video - Thanks!
The sprinkler design also seems relevant. The turn/bend in the sprinkler is closer to the center of mass, than the end of the suction/blowing arm. Wouldn't that mean that applying the same force to both the bend and the end will result in the sprinkler moving forward (opposite of blowing direction)?
There's a clue in "Surely You're Joking, Mr. Feynman": some twist in the test apparatus was observed, leading to RF's increasing the pressure to observe a larger effect (before shattering the test vessel). So, taking the book literally, we can discard the "no rotation" hypothesis. The direction of rotation was not disclosed, however - a bit of Feynman's mischief on display!
Have they tried adjusting the geometry to change the direction of the inflow coming in the center to induce rotation in the opposite direction, or even attempt to get close enough to balanced for the result to be chaotic, changing randomly each time the experiment is run?
Fantastic problem; interesting investigation. I have a problem, for wich I had no luck searching for a good explanation: forces in vacuum. You get a suction cup, press against the well finished surface of something heavy, make vacuum ...and it can be moved upwards. Or remember the historical event at Magdeburg, with horses playing tug of war. Wich forces are present, with such a strong value? Intramolecular?
Question. The area in the middle of the sprinkler was quite big for this experiment. In a regular sprinkler the area where the water would meet would be way more narrow so would that force still be strong enough to spin the sprinkler in reverse?
But the rotation caused by the turbulence inside the sprinkler doesn't dismisses that it may also rotate this way due to the low pressure at the tip of the exit holes as we can observe in the PIV. Can't the rotation be caused by both effects?
Also in 14:14 you can see that the suction of the water causes the water on the outer walls of the tubes is flowing towards the opening, applying friction that might cause a rotation (or cancel the counter clockwise rotation) clockwise. Every single effect of these should be looked into, every effect we can try to calculate it's contribution to the rotation.
What if you didn't have a big tube in the middle, that leaves space for vortexes to build up, but instead the two suction tubes didn't meet at all, and were both connected to a pump? So instead of one big tube going down, we'd have two tubes next to each other going down. Would that cancel out the vortex effect or would it still take place inside of each respective tube? And if the vortex effect wouldn't take place, then would the tubes still spin?
The way I see it, when you pump water out a sprinkler, it has momentum as it exists, giving ratational thrust. But conversly as water enters its coming from all kinds of different directions and so the force cancels out, and it stays relatively still, though it may move as its waterlogged initially.
So, basically the reason it spins backwards is the same reason the outside wheel of an axle in a turn spins faster. The particles begin at the same point and end at the same point; however the ones on the outside of the tube are moving with more velocity. Because of the way the tubes are lined up in this experiment, the higher velocity particles are hitting the lower velocity particles from the opposite tube. This is what's creating the spin. I assumed the sprinkler heads would pull them selves in the opposite direction of flow, but my assumption was more based the velocity change between the particles being sucked in (outside the nozzle) vs the ones inside the nozzle. Similar to how compressors work. Interesting finding for sure.
I dont get why they debated this so long when Isaac Newton gave them the answer centuries ago with conservation of momentum. Since all of the fluid starts at rest, the motion of the sprinkler is a function of the angular momentum of the flow exiting it. If there is any swirl (or net angular momentum) remaining in the central pipe as flow is sucked through, the head would move. It could be either direction due to any asymmetries specific to how the test is designed. I think the purest form would be if they had stators (flow straighteners) in the central pipe in which case the head would not move at all. This phenomenon is super common place in hovering aircraft like the F-35 or the Harrier where the inlets usually point forward but they dont move forward since their exhaust is pointed directly downward.
I love that the result is 1 - what any rando would have expected 2 - due to forces that hadn't even occurred to Feynman et al.
If i understood correctly the momenty is caused by the acceleration on the outside of the pipe which creates the imbalance of the waterflrow that creates the difference in the vortexes. So if we change the pipes to equal length straight pipes, how would that affect the outcome at all?
The mess in the center due to the asymmetric flow caused by the bend wouldn't exist and no spinning while sucking.
So the only thing generating angular momentum is where the in leading tubes end and actually blow. I'm just wondering if and if so how much you could change results by alternating the direction/tilt of the ends INSIDE. In the experiment they are only slightly offset, what would happen if you bend them, could you generate more or less spin doing so?
So did the pump feed directly into the sprinkler tank, or was it isolated by another siphon that fed into the sprinkler tank? Wouldn't the vibrations/flow of the pumped water into the tank also potentially affect the sprinkler?
Not more than the overflow. But you have to eliminate or minimise the cross flow through the tank.
Hang on a minute - Is it dependent on which side of the equator we do the experiment, or if we do it on a retro-orbital body?
Do you think the vortex region asymmetry (not the number of regions) comes from imperfections in the manufacturing of the parts?
It's the varied velocity across the incoming flow
I'd say that it's totally dependant on the shape of the sprinkler head & it's connected tubes. The minimal "S tube" type could still rotate forward, if the inner ends were vertical? However I'd expect a type such as a large cross tube with several smaller nozzles along it to rotate in reverse, due to the water being removed from the nozzle side & from flow to replace that from around the back of the tube?
In my experience suction force somehow gets negated much faster than a blowing force. You can easily notice this if you have a leafblower and an industrial vacuum cleaner (maybe that latter one only at work). You will never get pulled forward by suction, but you can easily feel the blowing force on your machine. My idea is that the counter-force of suction mostly gets negated by the environment, before directly interacting with your machine, while the counter-force of blowing is largely applied to the apparatus.
Subscribed. What a great video. I guessed it would not move at all. I did not see Dean flow coming as the solution.
It would be interesting to see them tweak the angle at which the tubes join the center to either balance or accentuate the difference in vortex sizes… just to show that it affects the turning rate. It would also be interesting to move the be de of the tubes further from the center to allow the difference in fluid velocity across the cross-section of the tube to dissipate, which would be expected to lessen the effect of the turning of the sprinkler. Alternatively, they could insert some kind of mixing vane inside the tubes to “scramble” the various velocities.