This is really cool. Some thoughts: * Your airflow is subsonic, so a converging rear cone will convert pressure at the back of your fan into velocity. Diverging will convert velocity into pressure. Your exit total pressure (static+dynamic) will equal your ambient pressure. I'd try a converging cone which results in the same area as your fan annulus. * You really, really need a full, round nosecap that diverts air entirely over the motor, and a tail cone as well. Pressure behind the fan will be higher than pressure forward of the fan, so you actually want to have a cooling air intake behind the fan, airflow inside the motor will be forward, and then have that airflow come out through a duct in the nosecone that ejects backward. I know, crazy. Try looking at the airflow over the F-35 DSI inlet bump. Air is weird. * Increase the radius of that inlet lip. I wasn't able to see, but you should have the lip turn a full 90 degrees, maybe you have this already. The inlet is producing thrust, since the airflow around it is lower pressure than ambient. Bigger = more thrust. * You don't mention, but you can determine the blade angle from axial speed vs tangential speed. * It looks like you are using flat plate airfoils. You should be able to do better with actual airfoils. You can look these up on m-selig.ae.illinois.edu/ads/coord_database.html. Use xfoil to run simulations to see how they work, and to adjust them. I'd try an SD6060 airfoil with camber around 4%. Try converting the stator blades to airfoils first, that might be easier. What CAD system are you using? Getting the airfoil spline from xfoil into CAD sometimes takes a few tricks. Final thought: I really like your test stand. Have you given any thought to * nacelle design? * testing in ambient airflow?
Hey, thank you so much for writing this up, it’s insanely helpful. All of it makes sense. I’m just realizing now that most OEM EDFs have the motor tail cones. Makes sense aerodynamically, never thought to add them. They also generally have noses that cover the whole motor area like you said. I do have an arbitrarily shaped airfoil profile, it didn’t come out great on the print though. I’m using SOLIDWORKS. I should probably use one of the NACA shapes though. I’d probably just manually trace over the contour. What do you mean by ambient airflow testing? Like without a lip?
@@mach10point4 I mean testing in a breeze! The ducted fan will work differently when it's moving relative to the ambient air. I used SolidWorks too, and have a workflow defined for getting airfoils into it if you're interested.
Great write up. Thanks for the explanation - Would this work the same way in water if I wanted to create an underwater thruster? How different would the aerodynamics vary from the fluid dynamics?
@3DProspecting Yes and no? Basically your question doesn't make a lot of sense. Aerodynamics is fluid dynamics. Hydrodynamics is also fluid dynamics. Aerodynamics is the study of air and other gases in motion. Hydrodynamics is the study of liquids in motion. But they are both fluid dynamics, which is the study of fluids in motion. As such a lot of the underlying principles and equations are the same. The reason we tend to break it up into two different fields though is that liquids and gasses tend to have very different reynolds numbers (the ratio of the inertial and viscous forces.) As such they are going to behave very very differently and efficient designs for one will not be efficient for the other. If you want to design something to propel something through water I would recommend starting by looking at the common designs for hydrofoils and the propellers used to push watercraft, and then deciding on the pros and cons for each of the current popular methods for propulsion while comparing that to your use case. That will probably get you closer, quicker, than trying to apply aerodynamic designs to a hydrodynamic situation.
Consider cranking the .stl export resolution in your design software WAY up, to avoid that "~90 sided polygon" profile (did a very rough count of 1/4 of the circumference) you're ending up with on the inside of the duct. That will be adding a lot of turbulence and potential vibrations and standing waves the system, as well as creating ~91 areas where the fin to duct gap is wider. You can see the significance of this at 4:00 where the blade has left many evenly spaced marks on the duct (that's how I counted the number of flats). Even better, you could use a piece of PVC tube or similar as an insert, or as the entire linear section of the assembly. That way you could avoid the rough surface caused by the layer lines, although I'm unsure of the concentricity tolerances of domestic grade tubes. It'll be much easier to fill and taper a single step between PVC and printed sections, rather than trying to sand the area inline with the propellor evenly without causing any variations in cross section. Alternatively you could print in a solvent vapour polishable filament such as ASA or ABS and allow a brief exposure to "acetone fog" to do the hard work for you. That method also increases the part strength slightly, by blending/unifying the outer layers of the shell.
Getting the airflow "straight" is pretty important, thats why the inlet helped. Stators (motor mounts) shaped as fins will help stabilize turbulent flow through the duct and exit the exhaust smoothly. Get a nose cone that covers the whole hub as well as a tail cone (as previously mentioned), this will help your efficiency a ton.
I noticed something on your last test, which is that the bolts you were using to balance the test stand shifted forward. It's counterintuitive, but I think this could actually affect the results slightly--if we think of the foot on the scale as one pivot point of the arm, then the distance the bolts sit from directly above that spot applies a torque to the arm, which has to be countered by a downward force from the books; as the bolts move, that torque increases, and thus slightly more down force is indirectly shifted from the books to the scale. This may not be significant, but grams are small enough measurements that I would try to eliminate this as a source of error, especially since it seems pretty trivial to accomplish.
To remove layer lines on PLA you will be better served by scraping rather than sanding. I use some cheap metal kidneys and they work great to smooth PLA.
I'm not surprised that you got such a boost from the inlet lip - if you look at commercial jet engines, they put a LOT of thought and careful design into their inlets - there's a reason theirs are extremely smooth, large and usually fairly rounded along the leading edge around the inlet. It helps smooth out the airflow going in, and essentially increases the angle of peripheral intake (so not just reliant on air coming straight in from directly in front). You might see a small increase in thrust by also changing the shape of the fan blade nose cone, as well as the struts holding the motor mounts in place - especially the leading edge of the struts would likely benefit from being rounded rather than flat (more shaped like the cross-section of an aircraft tail).
I mount my motors on the stator. I also use very slim inrunner brushless motors that take up much less space in the duct which produces more flow. The best EDFs use inrunner BL motors. There is no need for the torque of an outrunner motor for an EDF. It's wasting volume that can be converted into extra flow.
Something that's helped me a lot - I switched to a PEI coated spring steel plate from Amazon, and it's been amazing ever since. You could probably get away from all that brim very easily, while also having super easy removal from the bed when it cools. Quick note, if you begin having issues with adhesion suddenly on PEI, try washing it with hot water and dish soap, hand oils heavily affect the performance.
Yeah my plan is to switch to PEI asap when the glass fails. Which might be a while lol. As for the brim I think I was just experimenting I have no clue what the reasoning was
@@mach10point4 I'll help you get past that reasoning: the glass is "failing" every single time it requires you to use a brim on a part that shouldn't need it. Even if you only value your time as = minimum wage, it wouldn't take long for a PEI bed to "pay for itself" in de-brim/postprocessing time! Plus your part edges will be cleaner and parts more symetrically balanced, so are likely to give better performance and last longer!
Well done. I did some experiments with ducted props recently for a quadcopter. The motor mount is an obstruction to the outflowing air. I noticed flipping over the propeller and reversing the flow of air through the tube produced maybe 10% more thrust in my tests. The air coming in is moving relatively slowly and doesn't mind a small obstruction from the motor mount. Also, you can keep enlarging the input lip to increase thrust, just give it a parabolic profile. There are scholarly articles on this topic that aren't too hard to read.
As a retired blade prop manufacturer; the reason you are having vibrations on all of your blade designs is that you need to do one simple thing that I didn’t see in your project, and that is to calibrate the CG of your prop designs. Take for example a wheel, at high velocity without being properly calibrated it’s going to be unstable. Keep going and wear eye protection when doing these test. Good luck!
PLA for duct printed horizontally (not vertically because this way you're bound to get cracks & separation) and MJF for the impellers following a good old balancing works the best. I printed it during lovkdown, it still works with no issues so far. Its on my shorts, you can have a look.
This is a really cool project and you're probably learning a ton from it. I wonder if some odder fin shapes would change the sound profile while keep the lift, Zipline has some odd looking fins inspired by shapes in nature that makes their drones super quiet (Mark Rober has a video about it in the past year).
If the goal isnt to make it purely from 3d printed parts but to use 3d printing, you might be able to print an inverted mold to cast some resin to make stronger more reliable ducted fan. If you do do this, make sure to lubricate the surfaces of the mold with some kind of oil or wax so that the resin doesn't just stick to the inverted mold. Another thing which is handy for lightweight but strong components is making a 2 piece block press mold which you can place carbon fibre sheets and something to stick them together and press it into shape. If you are doing that for a system that has an afterburner or a combustion system, then use fireplace mortar as the binder between carbon fibre sheets.
If you are making a propulstion system for rc aircraft or for concept, you might also be able to print a compressor fan, as you can utilise the high base rpm of hobby motors and you can also use some servos at the rear with some rounded panels to make a controllable variable area nozzle for higher and lower thrust.
Nice work! I think it would be advantageous to make the propeller hub, motor mount base and cone the same diameter as the motor, as few flow transitions that doesn't help the flow/thrust as possible , and the transitions you can't avoid should be as smooth as possible. I maintain that an exhaust cone should in principle not help an EDF much, the fact that it does in this case suggests the design produce "extra" pressure, at the of cost lower potential flow, without the cone and or potentially giving opportunity for turbulent flow to unify slightly. To clarify, I don't think "forcing" the flow from a fan to increase velocity can result in net thrust gain worth the cone. Converging the flow to approximately the same flow area as around the motor, preferably in combination with a tail cone on the motor, and any rectifying of the flow it contributes to can help. It's whole different story with combustion engines, where the amount of exhaust is largely determined by the amount of fuel you manage to burn. While the motor mount "arms" was improved, I also till think they ideally should be shaped as airfoils, to act as stators, and not create as much turbulence.
Spot on, these are the primary things I'm looking at for the next iteration. Larger nose, motor mount nozzle, angled airfoil stators. And yeah, the nozzle theory makes sense to me, I'm just getting really weird numbers with the thrust tubes, and it varies wildly for each propeller/duct set up. Going to be a lot of trial and error for mow.
The thrust tube serves to help direct the flow of air, so that is coming out more straight. Otherwise it still has a lot of rotational motion, and instead of coming out straight, its coming out in a cone, which is a lot of loss. Perhaps could add some sort of stator (the motor attachment is supported by five fins that already do that just poorly) or a counter rotating turbine, like in jet engines
@@mach10point4 No, its not interchangeable with a stator. Stator helps straighten out airflow, baffle plate seems to slow it down instead, but i might be wrong on the baffle plate
Great video and setup! While your way ahead of me on this subject I do fly electric ducted fans, and in regards to motor type, they all now seem to use the outrunners, can't remember last time I seen an inrunner used. The outgunned seem to have the rpm and torque. Keep up the great work! Will be watching for your next video! Thanks for sharing w us!
This is super helpful, I'm just looking into motor types recently and yeah outrunners seem to be the move. I think I can comfortably handle a lot more rpm. Means a lot!
Consider using simple syrup for bed adhesion 1 part sugar, 1 part water. The print will self release when the bed cools often with a "tink" when it lets go.
what about adding even more, thinner, taller struts to hold the motor, make it act as a stator. add a slight twist to them so that the rotating air coming out of the fan is straightened out, so that all the velocity is on-axis. the twist direction is a bit counter-intuitive but i designed a desk fan where there was a huge difference when I got the stator twist just right
I tried that, but the plastic stators added too much drag. I also made a bunch of thinner ones from an aluminum can, and they did improve thrust a little but I decided it wasn't worth the work for maybe 5% additional thrust.
diverging duct won't increase thrust until the flow at the converging nozzle reaches Mach 1, otherwise it will decrease it. You can try tuning the geometry of the converging nozzle (especially the outlet diameter) which leads to a significant impact on thrust. You also can try outlet guiding vanes which increases overall drag but with bersize tuning of its angle wrt the flow, it converts all the swirls into a one-directional mainstream flow inharintly increasing thrust and eliminating moment.
If you ever consider printing ABS, try HIPS instead. Both will give you styrene poisoning, but HIPS emits an order of magnitude less. HIPS warps much less during printing than both ABS and ASA, is not prone to splitting, and is just much less rude. You can brush acetone onto it, it won't absorb deep into the surface but it can help smooth it out, though that's a little hit or miss, it's really much more resistant to acetone than ABS. It is very easy to sand, accepts paint and adhesives very well, and is very lightweight. It's also same price or cheaper than ABS. It is more chemically stable at elevated temperature, so it doesn't degrade in the nozzle during printing, so you can print it slow if you have to, which you can't do with ABS. It prints with a silky matte-ish surface.
This is a big help, I’d prefer ABS right now but I’m holding back because I don’t have a way to properly ventilate at the moment, so I’ll look in HIPS for sure
Would changing the motor mounts to have an angle to catch the air potentially increase the velocity? Making them act like the stators on a gas turbine so the air is smoothly transferred from a rotating airflow to a parallel one?
Im not sure you could get it printing on your printer but polypropylene filament prints with practically no layer lines it may be cheaper but take more time than a resin printer for smoothness
Great video. Thanks. How does your 3D printed EDF compares to the off the shelf commercial EDF? Did you manage to improve on the commercial ones? If I may suggest one minor change, is to 3D print motor support vanes at an angle, in order to further reduce friction. As fan rotavates it throws air tangentially, so air inside the tube flows in a corkscrew fashion. If you print motor support vanes straight they collided with corkscrew motion of air and cause drag.
Thanks for the tip! It's so hard to visualize flow without CFD. Right now, my 3D printed one is operating at about 1/3 the max thrust of off the shelf ones. I'm making a final video that goes over some slight improvements but more importantly the overall limitations of FDM.
Interesting. Could you do some testing on deeper blades? As in longer front to back, axially. Your current designs look like thin discs. Are you planning on sharing files in the future? I have future plans for testing stacked EDFs with a similar thrust testing rig.
Thanks for the tip. Been wondering where to go in terms of blade designs. Anyways, I will likely be posting all the STLs after my next iteration (including the test lever)
Why don't you use an intake shroud that is less diameter than blades? To reduce blow back from exaust. Then you can dramatically increase the converging nozzle to get super high exhaust pressure and velocity..
My understanding is that a fan inherently draws in ambient air outside of the fan radius, as opposed to an intake moving through the flow without a suction component. Therefore, a wider intake guides the flow that’s already being sucked in
That’s a good question, I have no clue. Someone else mentioned the rounded edges on jet engine nacelles. I suspect smooth surface geometry is probably more of a factor and intake area is less when the EDF is moving.
Very nice work. From the image on the video at the end, you can tell the housing could be better. You have probably printed a benchy. Make sure you perfect the benchy before getting four final numbers. You also talk about sanding, and people have mentioned acetone smoothing and not pla but resin printing which is also very nice. But those are final steps or post processing of pla. I think you can improve your print process with the printer you have. You could probably improve alot through working with layer height and print speed. You should think about nozzle thickness when working with layer height. Printer speed will also cause vibrations that make your print look like what is in the image. The Content of the video is very good. Also, for final results, 400g of thrust is probably very good, but I think you also made a mistake in your process of making too many changes at one time. For instance, changing blade type and adding that nozzle intake at the same time. I would say one change at a time. A plain housing until you find your best blade design. No converging outlet. This would isolate blade design. Then work with the housing, thickness, initial intake side shape, then exhaust shape, convergent design isolation. You then might have to look at blades or props again. Also the test bench is good. But you dodnt talk about whether or not the part that supports the motor and housing is the same length as the lever that acts on the scale. I think you should use a 1-1 so there is no torque multiplication. Maybe a more accurate reading if that is not what you did. In the video you cant tell, so maybe it is the same. That would be alot of work, but its a way to create more content. I think this was a good video and maybe you will make more or hopefully get even better results. Good luck!
I think your thrust measurements are off. The distance from the 3 hole fan mount to the axle, and axle to the center of the scale arm may be the same, but the fan itself then extends above the fan mount giving the fan a mechanical advantage, and skewing the thrust numbers you are getting higher than actual. If your only concern is comparing one design to the next, then it doesn't matter. If you want accurate numbers, then the upper arm would be reduced by half the fan diameter. (Center of fan to axle = axle to scale arm )
I *think* I designed it like that but I don't have the stand with me at the moment so I'm not sure. But that makes a lot of sense. For now more concerned with relative performance like you said, I'll definitely remake if it's off
Hey dude 😎👍 like the project only advice I can offer is to get a pull scale more accurate reading 🧐 you Could be robbing yourself of vital grams, it's also important to calibrate the back draft, you can calibrate this by adjusting the space between the fan and the inlet rim, test concurrently each millimetre you adjust monitor the thrust level & rpm calibration should allow rpm to peak successfully & not creating back draft at the inlet this is crucial for flow and at higher rpm can( withhold) thrust, if you are working with 11 blade fan my suggestion would be to place more stators than your blade count plus the degree of the stators should undergo multiple testing testing different degrees is very important, what may work on other commercial or home made EDFs may not be right for your project as each jet is different especially on dimensions, external inlet stators can assist with this to I hope I helped in some way 😎👌
You might want to trouble shoot your rig, it looks to me that the increase in thrust performance from the big nozzle may be an error caused by the shift in the centre of gravity over the lever arm. Just a guess. You could test that pretty easily by sticking a mass out the back and see if the performance changes. Or not.
you need Higher finns / Propellers. Just Look at Server Fans. There have much thrust and Sometimes 2 stacked. Also the Output hast some counter clocked static finns. So maybe U can do Something with it.
Keep it up chief - it's helping me get the motivation to keep with with fluid dynamics this sem If you do ever go for chemical smoothing; avoid Dihydrofuran and DCM unless you've got a fume hood around Ethyl Acetate is a much safer option, similar MSDS risk to acetone (so don't huff it) - though organic solvents are organic solvents; be careful if you do
Means a lot man, there’s probably a love hate relationship with thermo/fluids right now, I’m right there with you haha. And thanks for the tip, yeah it sounds like I’d have to go the super strong chemical route which I’m lowkey scared of so I probably will just have to go with an ABS/resin solution
@@mach10point4An acrylic floor polish like Future does some smoothing. It won't eliminate layer lines but fills in the sharp bottoms of the lines and makes a glass shiny surface. If it's too thick, thin it with some water. After it dries it can be removed with 91% rubbing alcohol.
Hi Jon！ Your work has really caught our eye, and we’re excited about the possibility of collabwith you. If you’re interested, please let us know the best way to reach out.☺
This is a pretty slick idea! If you are looking for metal 3D printed parts, please reach out! We are a startup offering affordable metal 3D printing services!
1. Make a proper inlet. Why only a tiny stubby lip? Make a roughly 30mm long parabolic inlet that opens up to 80mm, then add a full 180° lip with a radius of at least 10% fan diameter (1/4 fan dia long piece of straight shroud plus 1/2 fan dia long "inlet trumpet" always works well for bench testing). Inlet lip radius is critical and bigger is better, it's upper limit is more or less only determined by practicality. -> ruclips.net/video/OxKXHAQ7oXk/видео.html 2. That exhaust stator is an insult to any ducted fan and jet engine in existence. Just do something about it, it can't get any worse than those four bricks holding the motor right now. Probably an additional 5% thrust in that detail alone. (maybe try and copy something actually good, like a Schübeler fan, not $30 units from ali express that were copied and simplified three times already to be cheap, not engineered to be good edfs) 3. If static thrust is what you want, straight up double the current number of blades. You want the blades to overlap. The more you make them overlap, the more static thrust you'll get, but the fan will also get overloaded easier when you go into a dive to build speed. Tweaking the inlet cross-section and blade count for a good compromise will be necessary at some point. 4. A fixed inlet cone with guide vanes could add a good amount of efficiency (it also adds weight and is only useful at high airspeeds ... but the jet engine axial compressor look will be awesome when you'll eventually rip the fan out of the model and put it on a shelf ^^)
SOME FEEDBACK I WOULD SAY, CONCAVE THE PROPELLER A LITTLE FOR WEIGHT REDUCTION AND LESS CONTACT AREA. THEN ON THE HOUSING SUPPORTS, ANGLE THEM WITH THE ASME DIRECTION OF THE PROPELLER BLADES VERY SLIGHTLY. AND BRUSH SOME ACETONE ON THE PROPELLER SHOULD SMOOTH IT ALL OUT SO IT HAS SMOOTH SURFACE. WATCH A VID ON THAT BEFORE ATTEMPTING
What kind of thrust could you get with a dual prop edf? Where one is spinning one direction and the other the opposite direction. While producing the same directional thrust inline.
Really good question. Here's a paper I came across a few months ago apps.dtic.mil/sti/citations/trecms/AD1200423 that does exactly what you're talking about. I think the high level general consensus is that it helps a lot but is generally not worth the weight and energy penalty of the extra motor. I've always wanted to just try for kicks though
That should counter-balance the two rotors' rotational torques, and result in better control. It may help to have static blades to stabilise the airflow between the two fans.
This is really cool. Some thoughts:
* Your airflow is subsonic, so a converging rear cone will convert pressure at the back of your fan into velocity. Diverging will convert velocity into pressure. Your exit total pressure (static+dynamic) will equal your ambient pressure. I'd try a converging cone which results in the same area as your fan annulus.
* You really, really need a full, round nosecap that diverts air entirely over the motor, and a tail cone as well. Pressure behind the fan will be higher than pressure forward of the fan, so you actually want to have a cooling air intake behind the fan, airflow inside the motor will be forward, and then have that airflow come out through a duct in the nosecone that ejects backward. I know, crazy. Try looking at the airflow over the F-35 DSI inlet bump. Air is weird.
* Increase the radius of that inlet lip. I wasn't able to see, but you should have the lip turn a full 90 degrees, maybe you have this already. The inlet is producing thrust, since the airflow around it is lower pressure than ambient. Bigger = more thrust.
* You don't mention, but you can determine the blade angle from axial speed vs tangential speed.
* It looks like you are using flat plate airfoils. You should be able to do better with actual airfoils. You can look these up on m-selig.ae.illinois.edu/ads/coord_database.html. Use xfoil to run simulations to see how they work, and to adjust them. I'd try an SD6060 airfoil with camber around 4%. Try converting the stator blades to airfoils first, that might be easier.
What CAD system are you using? Getting the airfoil spline from xfoil into CAD sometimes takes a few tricks.
Final thought: I really like your test stand. Have you given any thought to
* nacelle design?
* testing in ambient airflow?
Hey, thank you so much for writing this up, it’s insanely helpful. All of it makes sense.
I’m just realizing now that most OEM EDFs have the motor tail cones. Makes sense aerodynamically, never thought to add them. They also generally have noses that cover the whole motor area like you said.
I do have an arbitrarily shaped airfoil profile, it didn’t come out great on the print though. I’m using SOLIDWORKS. I should probably use one of the NACA shapes though. I’d probably just manually trace over the contour.
What do you mean by ambient airflow testing? Like without a lip?
@@mach10point4 I mean testing in a breeze! The ducted fan will work differently when it's moving relative to the ambient air.
I used SolidWorks too, and have a workflow defined for getting airfoils into it if you're interested.
@@IainMcClatchie Ah gotcha. I'll let you know. Thank you again!
Great write up. Thanks for the explanation - Would this work the same way in water if I wanted to create an underwater thruster? How different would the aerodynamics vary from the fluid dynamics?
@3DProspecting Yes and no?
Basically your question doesn't make a lot of sense.
Aerodynamics is fluid dynamics.
Hydrodynamics is also fluid dynamics.
Aerodynamics is the study of air and other gases in motion.
Hydrodynamics is the study of liquids in motion.
But they are both fluid dynamics, which is the study of fluids in motion.
As such a lot of the underlying principles and equations are the same.
The reason we tend to break it up into two different fields though is that liquids and gasses tend to have very different reynolds numbers (the ratio of the inertial and viscous forces.) As such they are going to behave very very differently and efficient designs for one will not be efficient for the other.
If you want to design something to propel something through water I would recommend starting by looking at the common designs for hydrofoils and the propellers used to push watercraft, and then deciding on the pros and cons for each of the current popular methods for propulsion while comparing that to your use case.
That will probably get you closer, quicker, than trying to apply aerodynamic designs to a hydrodynamic situation.
Consider cranking the .stl export resolution in your design software WAY up, to avoid that "~90 sided polygon" profile (did a very rough count of 1/4 of the circumference) you're ending up with on the inside of the duct. That will be adding a lot of turbulence and potential vibrations and standing waves the system, as well as creating ~91 areas where the fin to duct gap is wider. You can see the significance of this at 4:00 where the blade has left many evenly spaced marks on the duct (that's how I counted the number of flats).
Even better, you could use a piece of PVC tube or similar as an insert, or as the entire linear section of the assembly. That way you could avoid the rough surface caused by the layer lines, although I'm unsure of the concentricity tolerances of domestic grade tubes. It'll be much easier to fill and taper a single step between PVC and printed sections, rather than trying to sand the area inline with the propellor evenly without causing any variations in cross section.
Alternatively you could print in a solvent vapour polishable filament such as ASA or ABS and allow a brief exposure to "acetone fog" to do the hard work for you. That method also increases the part strength slightly, by blending/unifying the outer layers of the shell.
Getting the airflow "straight" is pretty important, thats why the inlet helped. Stators (motor mounts) shaped as fins will help stabilize turbulent flow through the duct and exit the exhaust smoothly.
Get a nose cone that covers the whole hub as well as a tail cone (as previously mentioned), this will help your efficiency a ton.
I noticed something on your last test, which is that the bolts you were using to balance the test stand shifted forward. It's counterintuitive, but I think this could actually affect the results slightly--if we think of the foot on the scale as one pivot point of the arm, then the distance the bolts sit from directly above that spot applies a torque to the arm, which has to be countered by a downward force from the books; as the bolts move, that torque increases, and thus slightly more down force is indirectly shifted from the books to the scale. This may not be significant, but grams are small enough measurements that I would try to eliminate this as a source of error, especially since it seems pretty trivial to accomplish.
To remove layer lines on PLA you will be better served by scraping rather than sanding.
I use some cheap metal kidneys and they work great to smooth PLA.
I'm not surprised that you got such a boost from the inlet lip - if you look at commercial jet engines, they put a LOT of thought and careful design into their inlets - there's a reason theirs are extremely smooth, large and usually fairly rounded along the leading edge around the inlet. It helps smooth out the airflow going in, and essentially increases the angle of peripheral intake (so not just reliant on air coming straight in from directly in front). You might see a small increase in thrust by also changing the shape of the fan blade nose cone, as well as the struts holding the motor mounts in place - especially the leading edge of the struts would likely benefit from being rounded rather than flat (more shaped like the cross-section of an aircraft tail).
Great points. Filleting the inlet lips right now ✅
I mount my motors on the stator. I also use very slim inrunner brushless motors that take up much less space in the duct which produces more flow. The best EDFs use inrunner BL motors. There is no need for the torque of an outrunner motor for an EDF. It's wasting volume that can be converted into extra flow.
This is a great tip because I have been looking at other motors/KV ratings...it's all about speed for this one
Something that's helped me a lot - I switched to a PEI coated spring steel plate from Amazon, and it's been amazing ever since. You could probably get away from all that brim very easily, while also having super easy removal from the bed when it cools.
Quick note, if you begin having issues with adhesion suddenly on PEI, try washing it with hot water and dish soap, hand oils heavily affect the performance.
Yeah my plan is to switch to PEI asap when the glass fails. Which might be a while lol. As for the brim I think I was just experimenting I have no clue what the reasoning was
@@mach10point4 I'll help you get past that reasoning: the glass is "failing" every single time it requires you to use a brim on a part that shouldn't need it. Even if you only value your time as = minimum wage, it wouldn't take long for a PEI bed to "pay for itself" in de-brim/postprocessing time! Plus your part edges will be cleaner and parts more symetrically balanced, so are likely to give better performance and last longer!
Well done. I did some experiments with ducted props recently for a quadcopter. The motor mount is an obstruction to the outflowing air. I noticed flipping over the propeller and reversing the flow of air through the tube produced maybe 10% more thrust in my tests. The air coming in is moving relatively slowly and doesn't mind a small obstruction from the motor mount. Also, you can keep enlarging the input lip to increase thrust, just give it a parabolic profile. There are scholarly articles on this topic that aren't too hard to read.
Woah that's cool! btw you are SO under-rated.
🙃
As a retired blade prop manufacturer; the reason you are having vibrations on all of your blade designs is that you need to do one simple thing that I didn’t see in your project, and that is to calibrate the CG of your prop designs. Take for example a wheel, at high velocity without being properly calibrated it’s going to be unstable. Keep going and wear eye protection when doing these test. Good luck!
Great to hear from a pro. Thanks for the tip! And just got some testing glasses lol
DON'T STOP MAKING YOUR OWN THING WHEN IT'S AWESOME LIKE THAT
I might also try to fin shape the motor pylons with a sharp tip in front and behind so the air can flow around it easier.
PLA for duct printed horizontally (not vertically because this way you're bound to get cracks & separation) and MJF for the impellers following a good old balancing works the best. I printed it during lovkdown, it still works with no issues so far. Its on my shorts, you can have a look.
This is a really cool project and you're probably learning a ton from it. I wonder if some odder fin shapes would change the sound profile while keep the lift, Zipline has some odd looking fins inspired by shapes in nature that makes their drones super quiet (Mark Rober has a video about it in the past year).
If the goal isnt to make it purely from 3d printed parts but to use 3d printing, you might be able to print an inverted mold to cast some resin to make stronger more reliable ducted fan. If you do do this, make sure to lubricate the surfaces of the mold with some kind of oil or wax so that the resin doesn't just stick to the inverted mold. Another thing which is handy for lightweight but strong components is making a 2 piece block press mold which you can place carbon fibre sheets and something to stick them together and press it into shape. If you are doing that for a system that has an afterburner or a combustion system, then use fireplace mortar as the binder between carbon fibre sheets.
Yes I’ve been looking at molds but it’s seems very intimidating for DIY set ups, this helps me get started though so thank you
If you are making a propulstion system for rc aircraft or for concept, you might also be able to print a compressor fan, as you can utilise the high base rpm of hobby motors and you can also use some servos at the rear with some rounded panels to make a controllable variable area nozzle for higher and lower thrust.
Dope video! Keep it up the documentation and detail is great dude
Love your videos dude
Add a nose cone to the fan. And a much bigger velocity stack. You need a cleaner air inlet
Nice work! I think it would be advantageous to make the propeller hub, motor mount base and cone the same diameter as the motor, as few flow transitions that doesn't help the flow/thrust as possible , and the transitions you can't avoid should be as smooth as possible.
I maintain that an exhaust cone should in principle not help an EDF much, the fact that it does in this case suggests the design produce "extra" pressure, at the of cost lower potential flow, without the cone and or potentially giving opportunity for turbulent flow to unify slightly. To clarify, I don't think "forcing" the flow from a fan to increase velocity can result in net thrust gain worth the cone. Converging the flow to approximately the same flow area as around the motor, preferably in combination with a tail cone on the motor, and any rectifying of the flow it contributes to can help. It's whole different story with combustion engines, where the amount of exhaust is largely determined by the amount of fuel you manage to burn.
While the motor mount "arms" was improved, I also till think they ideally should be shaped as airfoils, to act as stators, and not create as much turbulence.
Spot on, these are the primary things I'm looking at for the next iteration. Larger nose, motor mount nozzle, angled airfoil stators. And yeah, the nozzle theory makes sense to me, I'm just getting really weird numbers with the thrust tubes, and it varies wildly for each propeller/duct set up. Going to be a lot of trial and error for mow.
The thrust tube serves to help direct the flow of air, so that is coming out more straight. Otherwise it still has a lot of rotational motion, and instead of coming out straight, its coming out in a cone, which is a lot of loss. Perhaps could add some sort of stator (the motor attachment is supported by five fins that already do that just poorly) or a counter rotating turbine, like in jet engines
Thanks. By stator, do you mean something like a baffle plate?
@@mach10point4 No, its not interchangeable with a stator. Stator helps straighten out airflow, baffle plate seems to slow it down instead, but i might be wrong on the baffle plate
Keep up the amazing work 😃❤
Great video and setup! While your way ahead of me on this subject I do fly electric ducted fans, and in regards to motor type, they all now seem to use the outrunners, can't remember last time I seen an inrunner used. The outgunned seem to have the rpm and torque. Keep up the great work! Will be watching for your next video! Thanks for sharing w us!
This is super helpful, I'm just looking into motor types recently and yeah outrunners seem to be the move. I think I can comfortably handle a lot more rpm. Means a lot!
Here's a recommendation: You should invest in a resin printer, they can produce much smoother parts for better flow characteristics.
Did you read the disclaimer at the literal very start of the video
Smooth isn’t good for airflow regardless what u think lol . Especially when it comes to thrust noise yes lol actually thrust no
Smoother is not better.
Layer lines shouldn't make tooooo much of a difference in thrust
Yah, there's benefits to resin, but it's not necessary by any stretch. I print with both, but honestly each has its purpose.
Consider using simple syrup for bed adhesion 1 part sugar, 1 part water. The print will self release when the bed cools often with a "tink" when it lets go.
Finish your parts by using weld-on #3 in a pre-val sprayer.
It actually bonds the layers together.
what about adding even more, thinner, taller struts to hold the motor, make it act as a stator. add a slight twist to them so that the rotating air coming out of the fan is straightened out, so that all the velocity is on-axis. the twist direction is a bit counter-intuitive but i designed a desk fan where there was a huge difference when I got the stator twist just right
I tried that, but the plastic stators added too much drag. I also made a bunch of thinner ones from an aluminum can, and they did improve thrust a little but I decided it wasn't worth the work for maybe 5% additional thrust.
Great accomplishment.
I haven't done much testing of stator like fan mounts, (similar to wind tunnel motor mount - stator combination) but it might increase your thrust.
Yeah I’m realizing there’s quite a bit of precedent for this stuff. Good to know
diverging duct won't increase thrust until the flow at the converging nozzle reaches Mach 1, otherwise it will decrease it.
You can try tuning the geometry of the converging nozzle (especially the outlet diameter) which leads to a significant impact on thrust. You also can try outlet guiding vanes which increases overall drag but with bersize tuning of its angle wrt the flow, it converts all the swirls into a one-directional mainstream flow inharintly increasing thrust and eliminating moment.
nice! it would be really nice to see how much thrust that same motor and battery can produce using a propeller
Yup! Great idea mind as well.
If you ever consider printing ABS, try HIPS instead.
Both will give you styrene poisoning, but HIPS emits an order of magnitude less.
HIPS warps much less during printing than both ABS and ASA, is not prone to splitting, and is just much less rude. You can brush acetone onto it, it won't absorb deep into the surface but it can help smooth it out, though that's a little hit or miss, it's really much more resistant to acetone than ABS. It is very easy to sand, accepts paint and adhesives very well, and is very lightweight. It's also same price or cheaper than ABS. It is more chemically stable at elevated temperature, so it doesn't degrade in the nozzle during printing, so you can print it slow if you have to, which you can't do with ABS. It prints with a silky matte-ish surface.
This is a big help, I’d prefer ABS right now but I’m holding back because I don’t have a way to properly ventilate at the moment, so I’ll look in HIPS for sure
Would changing the motor mounts to have an angle to catch the air potentially increase the velocity? Making them act like the stators on a gas turbine so the air is smoothly transferred from a rotating airflow to a parallel one?
Im not sure you could get it printing on your printer but polypropylene filament prints with practically no layer lines it may be cheaper but take more time than a resin printer for smoothness
Great video. Thanks.
How does your 3D printed EDF compares to the off the shelf commercial EDF? Did you manage to improve on the commercial ones?
If I may suggest one minor change, is to 3D print motor support vanes at an angle, in order to further reduce friction. As fan rotavates it throws air tangentially, so air inside the tube flows in a corkscrew fashion. If you print motor support vanes straight they collided with corkscrew motion of air and cause drag.
Thanks for the tip! It's so hard to visualize flow without CFD. Right now, my 3D printed one is operating at about 1/3 the max thrust of off the shelf ones. I'm making a final video that goes over some slight improvements but more importantly the overall limitations of FDM.
Interesting. Could you do some testing on deeper blades? As in longer front to back, axially. Your current designs look like thin discs. Are you planning on sharing files in the future? I have future plans for testing stacked EDFs with a similar thrust testing rig.
Thanks for the tip. Been wondering where to go in terms of blade designs. Anyways, I will likely be posting all the STLs after my next iteration (including the test lever)
Try angling the motor mounts into the exit vortices
I assume you will be testing a radius inlet ring for smoother inlet air and to increase forward pull from the front inlet radius
Yeah printing a massive one right now
Im trying the same thing, good to see im not the only one
Why would they not publish their max throttle thrust
Try using antislip pads or some rubber washer between the lever arm and the board to isolate vibrations.
Love this, doing this 100%
Cool! Do you think any resin/ metal parts may help for upcoming upgrade?
Why don't you use an intake shroud that is less diameter than blades? To reduce blow back from exaust. Then you can dramatically increase the converging nozzle to get super high exhaust pressure and velocity..
My understanding is that a fan inherently draws in ambient air outside of the fan radius, as opposed to an intake moving through the flow without a suction component. Therefore, a wider intake guides the flow that’s already being sucked in
how important is the inlet lip flare once on the platform in flight? as the flight path should push air into the fan
That’s a good question, I have no clue. Someone else mentioned the rounded edges on jet engine nacelles. I suspect smooth surface geometry is probably more of a factor and intake area is less when the EDF is moving.
3:37 why did you cut out the support removal
Printed on brim I have no clue why
Very nice work. From the image on the video at the end, you can tell the housing could be better.
You have probably printed a benchy. Make sure you perfect the benchy before getting four final numbers.
You also talk about sanding, and people have mentioned acetone smoothing and not pla but resin printing which is also very nice.
But those are final steps or post processing of pla.
I think you can improve your print process with the printer you have.
You could probably improve alot through working with layer height and print speed. You should think about nozzle thickness when working with layer height.
Printer speed will also cause vibrations that make your print look like what is in the image.
The Content of the video is very good.
Also, for final results, 400g of thrust is probably very good, but I think you also made a mistake in your process of making too many changes at one time.
For instance, changing blade type and adding that nozzle intake at the same time.
I would say one change at a time.
A plain housing until you find your best blade design. No converging outlet.
This would isolate blade design.
Then work with the housing, thickness, initial intake side shape, then exhaust shape, convergent design isolation.
You then might have to look at blades or props again.
Also the test bench is good. But you dodnt talk about whether or not the part that supports the motor and housing is the same length as the lever that acts on the scale. I think you should use a 1-1 so there is no torque multiplication. Maybe a more accurate reading if that is not what you did. In the video you cant tell, so maybe it is the same.
That would be alot of work, but its a way to create more content.
I think this was a good video and maybe you will make more or hopefully get even better results. Good luck!
Consider vapor smoothing your PLA as well
Very cool work. It's great that you iterate through all the designs. But can you simulate some of the iterations with CFD?
Running SOLIDWORKS off a student license and I tried but can’t get the flow tool :( would love to get one eventually of course
I think your thrust measurements are off. The distance from the 3 hole fan mount to the axle, and axle to the center of the scale arm may be the same, but the fan itself then extends above the fan mount giving the fan a mechanical advantage, and skewing the thrust numbers you are getting higher than actual. If your only concern is comparing one design to the next, then it doesn't matter. If you want accurate numbers, then the upper arm would be reduced by half the fan diameter. (Center of fan to axle = axle to scale arm )
I *think* I designed it like that but I don't have the stand with me at the moment so I'm not sure. But that makes a lot of sense. For now more concerned with relative performance like you said, I'll definitely remake if it's off
Print that thing at .08mm layer height using pc-cf (the fan not necessarily the housing
WOW, keep up the work. You should include the .stl's. You deserve more subs!
:D will post STLs sometime
Hey dude 😎👍 like the project only advice I can offer is to get a pull scale more accurate reading 🧐 you Could be robbing yourself of vital grams, it's also important to calibrate the back draft, you can calibrate this by adjusting the space between the fan and the inlet rim, test concurrently each millimetre you adjust monitor the thrust level & rpm calibration should allow rpm to peak successfully & not creating back draft at the inlet this is crucial for flow and at higher rpm can( withhold) thrust, if you are working with 11 blade fan my suggestion would be to place more stators than your blade count plus the degree of the stators should undergo multiple testing testing different degrees is very important, what may work on other commercial or home made EDFs may not be right for your project as each jet is different especially on dimensions, external inlet stators can assist with this to I hope I helped in some way 😎👌
You might want to trouble shoot your rig, it looks to me that the increase in thrust performance from the big nozzle may be an error caused by the shift in the centre of gravity over the lever arm. Just a guess. You could test that pretty easily by sticking a mass out the back and see if the performance changes. Or not.
hmm, I tare the set up every time so I assumed it wasn't an issue. Not quite sure if that's what you mean?
you need Higher finns / Propellers.
Just Look at Server Fans. There have much thrust and Sometimes 2 stacked. Also the Output hast some counter clocked static finns. So maybe U can do Something with it.
give some stl files
i want to make my own
Keep it up chief - it's helping me get the motivation to keep with with fluid dynamics this sem
If you do ever go for chemical smoothing; avoid Dihydrofuran and DCM unless you've got a fume hood around
Ethyl Acetate is a much safer option, similar MSDS risk to acetone (so don't huff it) - though organic solvents are organic solvents; be careful if you do
Means a lot man, there’s probably a love hate relationship with thermo/fluids right now, I’m right there with you haha.
And thanks for the tip, yeah it sounds like I’d have to go the super strong chemical route which I’m lowkey scared of so I probably will just have to go with an ABS/resin solution
@@mach10point4An acrylic floor polish like Future does some smoothing. It won't eliminate layer lines but fills in the sharp bottoms of the lines and makes a glass shiny surface. If it's too thick, thin it with some water. After it dries it can be removed with 91% rubbing alcohol.
Hey man send me a way to mail you and ill send you my EDF design and ill tell you how i print my EDF blades
Hey! My email is jonathanyun617@gmail.com
and.... subscribed.. by the way... look up Q-tipped propellers and see if that helps with anything.
Hi Jon！ Your work has really caught our eye, and we’re excited about the possibility of collabwith you. If you’re interested, please let us know the best way to reach out.☺
Gunpla sanding pads
This is a pretty slick idea! If you are looking for metal 3D printed parts, please reach out! We are a startup offering affordable metal 3D printing services!
1. Make a proper inlet. Why only a tiny stubby lip? Make a roughly 30mm long parabolic inlet that opens up to 80mm, then add a full 180° lip with a radius of at least 10% fan diameter
(1/4 fan dia long piece of straight shroud plus 1/2 fan dia long "inlet trumpet" always works well for bench testing).
Inlet lip radius is critical and bigger is better, it's upper limit is more or less only determined by practicality. -> ruclips.net/video/OxKXHAQ7oXk/видео.html
2. That exhaust stator is an insult to any ducted fan and jet engine in existence. Just do something about it, it can't get any worse than those four bricks holding the motor right now. Probably an additional 5% thrust in that detail alone. (maybe try and copy something actually good, like a Schübeler fan, not $30 units from ali express that were copied and simplified three times already to be cheap, not engineered to be good edfs)
3. If static thrust is what you want, straight up double the current number of blades. You want the blades to overlap. The more you make them overlap, the more static thrust you'll get, but the fan will also get overloaded easier when you go into a dive to build speed. Tweaking the inlet cross-section and blade count for a good compromise will be necessary at some point.
4. A fixed inlet cone with guide vanes could add a good amount of efficiency (it also adds weight and is only useful at high airspeeds ... but the jet engine axial compressor look will be awesome when you'll eventually rip the fan out of the model and put it on a shelf ^^)
Balancing seems to be poor in all iterations. Maybe you need a rig to balance each blade
A gram is a measure of mass, not force. Try millinewtons.
SOME FEEDBACK I WOULD SAY, CONCAVE THE PROPELLER A LITTLE FOR WEIGHT REDUCTION AND LESS CONTACT AREA. THEN ON THE HOUSING SUPPORTS, ANGLE THEM WITH THE ASME DIRECTION OF THE PROPELLER BLADES VERY SLIGHTLY. AND BRUSH SOME ACETONE ON THE PROPELLER SHOULD SMOOTH IT ALL OUT SO IT HAS SMOOTH SURFACE. WATCH A VID ON THAT BEFORE ATTEMPTING
Acetone doesn't work on PLA
AWE DANG NEVERMIND MY BAD@@greggv8
What kind of thrust could you get with a dual prop edf? Where one is spinning one direction and the other the opposite direction. While producing the same directional thrust inline.
Really good question. Here's a paper I came across a few months ago apps.dtic.mil/sti/citations/trecms/AD1200423 that does exactly what you're talking about. I think the high level general consensus is that it helps a lot but is generally not worth the weight and energy penalty of the extra motor. I've always wanted to just try for kicks though
That should counter-balance the two rotors' rotational torques, and result in better control.
It may help to have static blades to stabilise the airflow between the two fans.