As a 15 year old with the attention span of a chicken nugget, I though this video was gonna be boring asf because it's a 600 sub channel and an 18 minute long video. Dude, you did amazing! I sat though the whole thing and enjoyed it! Thank you so much :D If you're wondering why I clicked, I saw a plane with a mustache. How could I resist?

Dude, you put in so much work on your own; I can only imagine how much more awesome the GA community could be with more people like you improving the daily flight experience one step at a time. Amazing work!

I would suggest searching the KZfaqs for Mike Arnold videos addressing interference drag as well. He has passed but someone is doing a great job of archiving his recordings. In addition to increasing radius, experiment veritical placement in relation to the main wing ie: placement of max thickness of vertical compared to main wing, moving the location of the vertical fore and aft to see results. Try to keep the frontal area constant and see what you get.

Hey great investigation and video! I see that you've kinda struggled with getting good CFD data. I'm a PhD student studying turbulence modelling in CFD and if you're interested we could discuss some of your results and get your CFD game to the next level :) Wish you all the best and continue your awesome work!

Pretty cool. Wish we knew this in 83 when we built our eze. At that time the straightest build was the fastest due to having the less trim tab factored in to fix crooked builds. We used a builders sight to shoot straight lines and to get the eze as square as possible. 75% we ran at about 190 that was a stock veri eze, no starter, 100hp Lycoming with a prop we bought from Rutan him self. Out of 6 eze’s at the airport ours was the quickest. That radius curve at the winglets would have been interesting.

This man understands

This was awesome! I appreciate your intrepid masochism in learning that software and the charming quirks you encountered in them.

There are some other side-effect variables to consider.(Grow test-matrix, grow!) The way it appears that you created these curves is to basically cut the corner; which slightly reduces frontal area, it also reduces effective horizontal wing area(similar to the way it reduces effective vertical surface). the other effect is that at high AOA the curve tends to dumps air from the lower surface. Whether these side effects are significant and how they could be compensated through changes in cord and span, is the question of course. I know for a simple winglet, which is not a vertical stabilizer, a simple increase in wingspan is generally more efficient than a winglet of similar frontal area and mass. The exceptions are cases where the bending stress on the wing causes disproportionate added structural mass.(Which is all quite dependent on wing thickness, struts, intended mission, initial aspect ratio, and so on.)

This was really cool to watch. Documenting the whole process from start to finish was really something else. Nice worK!

Fantastic video, keep up the good work. It's super awesome to see that you even had to teach yourself CAD for this. It's always great to see someone go to such lengths to test their knowledge on their own, and even better to see them share those findings with others!

Wow, so much work done by yourself, and resulted in success! I admire you!

Very well done. Your time and effort is much appreciated.

Thank you so much for your impressive work on this question! it helped me a lot to understand the aerodynamic issues around winglets better!

Thank you for taking us on our journey and thinking process

Thank you for putting so much work into this project, I found it very interesting!

Wow! The quality and depth in this video is like that of a master's thesis. I can only imagine how much time and effort went into it, but I'm sure the skills you learned along the way will be useful for all sorts of other things down the line. And who knows - it might even save someone's life, knowing to avoid the potential stall situation. Nice work.

Insightful. Due to Mike Arnold's videos I knew how to decrease interference drag for low wing airplanes. You raise a valuable point for high wing airplanes too.

Bravo! A very helpful study. I have a new appreciation for how the slightest angle can make a difference. It’s important to do research. Thank you for doing this.!

And thank you for taking the time and effort to make this video and present your research! It has the same vibes as "Think Flight" videos. I really like this type of video. I remember reading or watching something from Mark Zeitlin where he presented his belief that a blended winglet would not appreciably benefit a Cozy since the Cozy already had a good design where the vertical leading edge began aft of the main wing's leading edge. That is possibly something to look into as it would probably be lighter and easier to build.

Clear, Concise and professional. Good Job.👍

I work on cars and I have to say I’ve been watching a lot of Aviation channels and this is by far the most detailed video I’ve seen on Aircrafts mainly one’s aerodynamic behaviour. My friend got me into planes and I want to learn how to fly let alone work on planes. This is the channel to grow my knowledge with no doubts.

Excellent job on the analysis!

Great video. Very easy to follow, even for a layman such as myself.

Love your use of Kerbal Space Program

Fantastic project and video!

Man, you are good, in your standing to get result, even if first long-term tries did not gave you the success.

Great analysis!

Brother 🙌🏻 one of the most useful and interesting videos I've ever come across while looking 👀 into aviation and Long EZ/ Cozy information. Thank you so much ✈️👮🏻‍♂️🛩️🇺🇸🛫

Great work bro

Out standing. I believe the canard plane has been attempted in various forms. Even a jet, perhaps. It does seem that for a 4 seater multi composite structure. Supporting a turbine would be doable if, similarly designed like the SF50 in turbine placement and tail design. Your video was spectacular. Please continue to share.

Super interesting. Great job.

fantastic work!!

Beautiful work, thanks bunches:)

Very good work!! My work in RF has me running lots of 3D electromagnetic simulations using Ansoft HFSS. Also I run multi-dimensional simulations of circuits using Keysight ADS and Cadence Spectre. So I totally get the trades between accuracy, simulation size, time, and tool capabilities - along with licensing restrictions. Kudos to you for persistence and for developing good insight. I think what happens a lot with running simulations is that one’s biological neural net gets trained. That training, I think, is a primary benefit of using simulators. They may not always provide accurate results. But they will definitely educate your mind about trends. A lot of times this training on trends and principles leads to innovation. Again, great work well explained.

This was really interesting👍

Very interesting - great vid!

On the Cessna wing-fuselage intersection (6:26), a small fairing to be attached to the rim of the door between the upper edge of the window and the top of the door. Whether that would be large enough to be significant is still a question. There would also be a small interference near the front edge of the door as it hinges out, but that could be minimized. Given the size of the wing strut fairing and landing gear strut fairing, and the gap between the fuselage and landing gear strut fairing, it seems like a small fillet fairing on the top of the door would still be helpful in reducing drag.

good work taking it from a suggested radius to showing a law of diminishing returns curve. i am sort of following the DBT Areo development work....that may be introduced as a kit at some point.

Very good informative video documentary....!

This is amazing! I flown a flying wing with both wingtip winglets and without and I noticed the winglets caused significant drag

Wow a really great video. I actually have an SQ2000 which originally had E-Racer style wings. I have since changed these for hand layup blended winglets built by Jack Morrison himself. I think they are the 9" radius. They were not only much lighter, but will have the drag improvements you report as well. Am looking forward to the finished aircraft. For photos and details feel free to PM me.

Congratulations

nice workup 🎉

Excellent, many thanks. Subbed.

It's when looking at these tools that I'm happy I used to work with converge science. It has adaptive resolution for simulation. So an areas with low compressibility and interest it only simulates it something like two to three inches whereas areas of interest it can simulate with resolutions finer than millimeters. It also has the ability to do moving surfaces. Requires a supercomputer to run and an expensive license but boy is it powerful. Still takes hours and hours to set up a run only for it to maybe fail halfway through but when it works it's amazing

Very interesting. I was wondering if you had also considered a couple other questions. What would happen if you canted the winglets out a small amount but used a smaller radius, I.E. put a 3-4 degree outward can’t but only used a 6” radius. Also in the winglets, if you moved them back slightly, if you look at the winglets of the Long EZ compared to the E-Racer you’ll notice that the Routan winglets actually trail behind the wing (most likely to get them in cleaner air and make the more effective) You’ve done excellent and very thorough analysis, thank you

Outstanding work presented, well done. A comment from ibonito seems to offer access to more powerful CFD which will improve your results which you clearly understand. If this is your master's thesis, I "grant" you that degree with high distinction.

Evidently, Van's has studied "Wing-Root" fairings to add a curve. They found that it actually caused more parasitic drag. Perhaps because the air over the fuselage starts well before the wing, that it is less important there. Awesome work on the wingtip though. On my Cessna T210, I have a STOL Leading Edge Cuff that also came with Down-Turned Winglets. So after viewing your video, I wondered if upward or downward would matter. Given that there is higher pressure below the wing, I would think that downward turning would be better (given ground clearance. It appears that studies to confirm that downward turning works better than upwards turning concerning parasitic drag only Great work, fascinating video

Nice work. There's still lots left on the table. The AoA of the winglet can be set to provide some net thrust because the induced flow at the winglet is no longer axial and so can provide enough of an angle to generate a lift vector that tilts forward enough to be useful. This is a pretty crap explanation but might help give you the idea. Also, you may find that a panel code such as is available in something like Aeolus or XFLR5 might be more useful for doing the bones of this type of analysis since it is much less computationally intensive and thus you can iterate many more times. Good luck with your investigations!

wow, great video

Great findings! Love it! The next step on a ERacer is to eliminate the canard configuration and therefore get a much smaller mainwing.

My God, I was already interested in planes when the VariEze was still new - such videos show me how old I am. I have seen a few numbers about winglets, and from what I seem the rise in efficiency the winglet brings is about the same when you would lengthen the wing by the length of the winglet, i.e. increase the aspect ratio. So that would be the better road to go, lengthen the wing. But many planes have other restrictions regarding wing length, and then the winglet is the right choice. Like on a passenger jet where the place (= width) on the airport terminal is restricted, or when you have a restriction like the 15 Meter-glider class.

You might also want to investigate prandtl style wingtips just inboard of the tip and prior to the winglet. This will locally reduce the pressure variance (local to the wingtip) between the upper and lower airfoil surfaces. This tip design could also incorporate winglets with less form drag.

Aesthetically more pleasing to the eyes without the front wings, plus extra advantage less drag.

Great video. As you mentioned you need high fidelity CFD simulations. OpenFOAM is a great option ( you didn't explain your problem with OF). There are other options too, such as SU2 or Lattice Boltzmann cfd codes. I'm learning shape optimization with DAfoam. I think you should give it a try

Intriguing

Wow, lots of work! Reducing the interference drag by blending the intersection to a generous radius is well proven in several blended winglet studies. Hoerner also deals with the subject empirically, if you consider the wing and winglet to be two struts. What would be interesting and arguably more useful to other builders is a structural analysis on the new winglet, and determine what radius is acceptable from a builders perspective. And how the new spar would be constructed.

On the 172 could a curved lip at the top of the door help the airflow between the wing and slab door while still allowing the door to open?

Interesting, thanks for sharing! Have you considered investigating the effects of moving the fins back a bit? Having to perpendicular surfaces with increasing pressure, i.e flow passed the thickest point, promotes separation. An example of solving this kind of separation, and associated drag, problem can be seen in designs where the fuselage max width is located aft of the wing. That reduces the need for fillets.

Its like a long EZ, but faster. My experience with a long EZ left a good impression, and it is a really nice and responsive plane

@walker Weathers damn man! Killed it on this, was wondering what if you put the rear canard on the top so almost like a cesna canard thing, give everyone epic views but could also add a while cargo area at the rear under where the engine could go. and do the winglets work in reverse so have them pointing down since the wing is now much higher?

at 1:34 there are some small winglets pointing straight down below the main wing. I remember reading somewhere that some builders omit these which can have some serious safety concerns in specific scenarios but I forget which. Anyway assuming he lower winglet is necessary, I wonder what the best bends would be for each, one curvy, the other straight? or both curvy? or both straight?

Fantastic! Now, - since you had to learn using the software specifically for this purpose - how about a compact tutorial that touches explicitly only basic modelling and setup of simulations?

I like that you even mentioned "diminishing returns", which are returns, nonetheless. Modern designs like the Boeing 787 wing, seem to indicate that the largest practical radius is optimal. A square has more perimeter than its inscribed circle. Besides reducing interference drag, increasing the winglet radius also reduces frontal area. Less frontal area means less drag. Structurally, it results in less mass, less cost in (composite) material, and the removal of a difficult-to-layup joint.

Good video

This is fascinating, I wonder if additional gains can be had with some canard fairings. Also makes me wonder how much a typical 172 has at the root

It sounds like a good improvement. One potential problem could be the rudder cable would somehow need to be raised to the new higher lower rudder location or some sort of extension of the rudder horn perhaps on a tube would have to be figured out.

You can try running the simulation with time steps and iterations. That would let the simulated flow stablize more quickly.

Your microphone sounds fantastic man lol. I prefer a more bass-heavy mic sound anyway

Great work and interesting! I don't think I am going to blend the winglets on my Cozy IV, it is fast enough :). I understand how much work that must have been. I tried to model a reverse scoop for my oil cooler use a freeware 2D CFD program. Gave up and just made a cardboard one and flew. My son did his senior engineering with CFD modeling. He had to design a cooler for a nuclear reactor. He had to use the schools super computer to do the model in a reasonable time without errors.

A fatal accident is reported to have been caused by flutter in a blended wing/stabiliser design. I was talking to an ez builder about your video and he told me that blended wing/stabilisers become very sensitive to flutter due to the decreased stiffness of the wing/stabiliser interface. He told me there was a fatal accident caused by this in a long-ez. I've asked him to share a report about the incident and will share it here if he is able to find it. You can easily test the inherent difference in stiffness with a folded paper model

That 12° AOA separation is interesting. Has anyone done active vortex generators for low speed flying? Or is that just generally not a thing? Also, I would be super fascinated to see a blended endplate or blended dual winglet version of this. I really need to get a laptop good enough to start doing CAD lol

Note to self: I've been putting off my experimental build for years - I better get going!

In case anybody is intrigued by this by this rear mounted wing and cannard design, you should look up the commercial failiure that was the Beechcraft Starship. Very interesting plane.

The detached flow you saw at 12 degrees aoa is not necessary a main wing stall. Since it was a small region near the back of the wing, and only a small spanwise section it may have just been local seperation, which the pilot would feel as buffeting.

Huh. I have a glider and retro fitted winglets. Drag is everything to us. I don't have to worry about vertical stabilizer (I still have a rudder). I'm probably going to review the angle blending my wing to the winglet. Thanks.

Did you/can you model a lover component to the winglet much like the 737-800+ tips? Cozy driver here. Very interesting. Good job 👍

Wow! Good job! I wonder if it’s worth using a Saber type winglet as seen on some 737s.

The design is actually the successor to Japan's WWII Mitsubishi Zero fighter called the Shinzen. The Shinzen prototype design which initially had an aft push prop but next version was with a primitive jet engine or rocket.

A point of pondering. In a box wing configuration where the verticle connecting Winget is also acting as a rudder, would both the upper and lower connection points benefit from a curve, or just the lower one?

The gains are real. That wing winglet intersection is a big problem and a nice sized fillet does wonders. I think it was probably worth 10kts on a long Ez. I recall tuft testing and seeing vertical flow aft the wing TE and up the rudder...and saw a bird literally get caught in the wake, bounce around for a few seconds and then get spit out.

Very good investigation. One thing I wonder is if you would have run those 12 degree simulations at a lower airspeed whether the flow would not have separated. 12 degrees AoA at the cruise speed would be a very high wing loading, not realistic for most stall situations which occur close to 1g at normal aircraft weight.

i wonder if the larger leading radius would work better if you were using a conical curve instead of a cylindrical curve, so that it tapers to a point at the trailing edge, giving you more vertical real estate for the vertical stabilizer. would be a harder shape to fabricate, but i would imagine it has better performance on paper.

Fascinating! But the Bearcat has no wing root fairings... is that similar?

Do you think that you're findings would apply to a blended winglet? I've noticed on the blended winglet that the upper winglet has a sharper (lower number) angle than the lower winglet. I ask, because I want to build composite wings for an RV-6, and incorporate a blended winglet. I would appreciate any input that you might have.

not an airplane designer... but could the smaller vertical stabilizer issue be fixed by switching from only having the trailing section of the stabilizer actuate to having the whole surface rotate? (IIRC the f16 uses this for its horizontal tail sections (but for different reasons ??))?

Outstanding! I've been pushing this for years, ever since working through some computer (Fortran) simulations in the 1980s. We examined winglets whose root chord equaled the wingtip's chord. As for the other parameters: Winglet tip chord: 0% (sharp point) to 200% of winglet root chord Winglet angle relative to horizontal: -120 degrees to 120 degrees Winglet angle of attack relative to winglet root (straight ahead): -30 degrees to +30 degrees Winglet twist relative to winglet angle of attack: -45 degrees to +45 degrees We were exploring extremes and quirks, and really wanted to know what might happen if the winglet were not only canted down, but so much that it was tucked under. Might their be benefits of exploring configurations well outside the boundaries of conventional wisdom, which suggest about a 60 degree angle up and about a 3 degree outward twist? We initially ran it as 1% and 1 degree increments. We quickly realized the PC could not handle 259,200,000 data points! So, we adjusted the increments to 10% and 10 degrees, 25,900 data points, and found no benefits of strange dimensions outside of conventional wisdom. We then re-ran the simulation using 1% and 1 degree increments to +/- 10 units either side of the identified peaks. So, another 160,000 data points. I do not recall the exact final figures, but it's of little consequence as far better simulations and testing have been done since. Besides, we were running this on IBM PCs, the original 4.77 MHz models, and with 640k RAM and 360k floppy drives, no less! I do, however, recall that it was around the following: Winglet tip chord: 42% of winglet root chord Winglet angle relative to horizontal: +47 degrees Winglet angle of attack relative to winglet root: -6 degrees Winglet twist relative to winglet angle of attack: -3 degrees We only explored the reduction in induced drag at POH-published standard-day cruising airspeed at mid-point of a leg of a distance commensurate with normal loading and enough fuel for 80% of MGTOW. Again, we developed the software and equations from earlier published models, so it was more of an exercise in how to conduct simulations than in finding the right answer. Over the years, however, I've been pleasantly surprised to find we were on the right track. I wish we could have have analyzed curved winglets, but we couldn't find any such published equations. Even if we had, it would have required vastly more processing power than we had at our disposal.

4:44 you might discover that lift does not wor the way you think it does: By "slicing" through the air, the wing accelerates the air downward actually using both surfaces of the wing. The acceleration of air "particles" along the flight path over the top surface of the wing is only a result of the general downward acceleration. From a static frame of reference the wing even accelerates the air forward due to the friction you mention. In the end, the idea derived from Bernoulli's principle that a wing creates lift from "underpressure" above its surface by accelerating the air above it against the direction of flight will not yield nearly enough lifting force. Wind channel observations do not necessarily provide all insight necessary to understand the total balance of (kinetic and pressure) energy states of the air surrounding the wing. 😊

The fairing on the Cessna could just as easily ave been part of the door if they truly wanted to get rid of interference drag at that point of the fuselage/wing junction

Very interesting. But in your description of lift from a wing, why did you not mention flow turning?

16:50 If the problem with increasing the radius is the loss of control surfaces, you should test the different lengths of rudder for a specific radius. For example, if you feel that a 12" radius starts to compromise the control surface too much, try 12" of radius, but bump the tip of the winglet up 3". Likewise, you could try 15" with +3" and +6" inches of overall winglet height. That would result in the same amount of control surface as with the 9" radius, which, in turn, might allow you to keep (most of) the L/D improvements of the larger radius, without compromising handling.

Or - move the vertical stabiliser backwards about 10 inches, to a foot (ish). Have a look at the current iterations of the wing-foils on the Americas cup yatchs if you dont understand what I mean. This seperates fore/aft the two areas of max interferance between the two foils (vertical and horizontal).

Hey, I saw the mesh you generated for your CFD simulations. Small problem with it, you didn't add any inflation layers (aka prism layers) to capture boundary layer behavior more accurately.

thx, man

You can try cloud GPU services for complex calculations, i.e. AWS NISP instance.

Hi, very interesting and helpful. What do you think or a wingtip root mounted propeller, mainly in the pusher configuration? Do you think this can reduce vortices? I understand that this is not able to provide a rudder effect but I am considering drone designs where the effect of a rudder comes from vectoring fans, these may even be on the wing tip. I am hoping that pusher configuration has the same effect as a puller configuration at reducing vortices but with laminar flow wing. In thinking about your design, I wonder if suction within the wing via a very slender fan and holes to suck boundary layer off could also eliminate interference drag at the winglet root.

Next go round suggest generalizing by expressing fillet radius as % of chord. (r/c). Good work. D

Wonder if you could run similar tests but to a boxed wing to see if there are any benefits to making a boxed wing aircraft

I wonder what results you would get with fences added to the underside of the wing. Perhaps the stall is partially due to vacuum created by the air on the underside running spanways toward the wing tips? Just a thought. People report better performance with the fences so maybe you should rerun your wing with them.

You may think I’m nutz, but I saw winglets on HUGE ceiling fans in the Martin Baker factory at Denver, many years before they appeared on aircraft wings