Experimental EDF Jets and Other Ideas

[L Edge]

It's something I've thought of looking at, but I'm a bit worried about damaging my EDF units. My understanding is that the cooling air flows through the EDF from the back to the front so whatever design I come up with would still have to supply the back of the EDF with high pressure air

Freewing 2836-3300Kv Brushless Outrunner Motor [MO028363] Motion RC

Freewing 2836-3300Kv Brushless Outrunner Motor - MO028363 This brushless outrunner motor is used primarily with Freewing 64mm 12-blade EDF jets with 4S power. It can be used with other aircraft with compatible specifications (see below). Features: Precision machined parts and high-quality...

www.motionrc.com

I have seen Freewing's motor's in the past on models being sold. I found one that has the cone at the end with holes. Now I know why they designed it differently than others. With lower pressure behind the motor than front, probably cools and produces additional thrust with cone shape. Or it flows from hole to center exit.

 

[telnar1236]

Stuck it in CFD, and you're not wrong about the improvements in efficiency. Losses are halved from around 14% without a tail cone to around 7% with the tail cone - I may need to look into it a bit more in the future. These are the results assuming a flow speed of 70 m/s through the fan. In this case, a more negative value on the fan side is better since the pressure is fixed at zero on the outlet side.

I also want to investigate the effects of a converging nozzle. The 14% losses line up very well with my measured values from the test with the 50mm fan and straight duct which is what I simulated as well so I'm thinking the combination of a tail cone and a duct could be even better.


Weirdly enough, the flow direction through most EDF motors is like this.

Due to the Bernoulli Effect, the pressure behind the motor is actually significantly higher than the air next to it since as the flow area rises, the speed falls, increasing the static pressure. This drives the air to flow in the back and then out the holes in the sides. I'd have to track them down, but there are videos out there with people using smoke to show this too. I expect centrifugal force on the air just after the fan has an effect too since because it's spinning, it wants to go outwards.

Also, the stream line tracker showing the vortex ring after the motor without the tail cone is one of the neater looking bits of CFD I've done in a while, so I figured I'd share it even if it isn't all that informative about the efficiency.

 

[L Edge]

Stuck it in CFD, and you're not wrong about the improvements in efficiency. Losses are halved from around 14% without a tail cone to around 7% with the tail cone - I may need to look into it a bit more in the future. These are the results assuming a flow speed of 70 m/s through the fan. In this case, a more negative value on the fan side is better since the pressure is fixed at zero on the outlet side.
View attachment 254376
View attachment 254377
I also want to investigate the effects of a converging nozzle. The 14% losses line up very well with my measured values from the test with the 50mm fan and straight duct which is what I simulated as well so I'm thinking the combination of a tail cone and a duct could be even better.


Weirdly enough, the flow direction through most EDF motors is like this.
View attachment 254378

Due to the Bernoulli Effect, the pressure behind the motor is actually significantly higher than the air next to it since as the flow area rises, the speed falls, increasing the static pressure. This drives the air to flow in the back and then out the holes in the sides. I'd have to track them down, but there are videos out there with people using smoke to show this too. I expect centrifugal force on the air just after the fan has an effect too since because it's spinning, it wants to go outwards.

Also, the stream line tracker showing the vortex ring after the motor without the tail cone is one of the neater looking bits of CFD I've done in a while, so I figured I'd share it even if it isn't all that informative about the efficiency.
View attachment 254379

That's why I asked to mount it on one pylon to reduce rotational flow.

 

[L Edge]

Here's one I explored flow wise that I called "thrusters". Wanted to hover and remain still vertically using a single EDF and no rudder. As you know with no airflow, torque from the fan wants to rotate it.

So here was my approach to solve it. Used a F-22 which has a exhaust duct and decided to counteract the torque by:
1) Using carbon fiber tubing with an inner opening of 3/16 inch so it traps airflow and travels down the tube.
2) The tube was placed about 1/16" perpendicular to the blades to pick up the total pressure on both sides of fan 180 degrees apart right at edge of fan.
3) Tubing is inside ducting and runs down to the CG line and makes 90 degree thru ducting and out to the edge of wing. This is done the same to the other wing.
4) Torque equal force times distance. At the end of the wing is a 90 fitting to counteract the torque of the EDF. So as the EDF flow comes out of the wing tubing perpendicular, there is a force. Get the distance to center of plane and you now have produce a counter-torque. Equal it hovers.

Testing was done in a dome. Used a TVN to take care of yaw and pitch. For roll, the thruster was marginal (time-30sec,1:48, 1;56 run at .25 speed) and works ok.

Had a 5 bladed 64mm EDF to use and I did use Scotch Magic Tape to reduce gap between blades and fan housing and with all that garbage tubing blocking and causing turbulent flow, still had enough thrust to fly around. I wonder if I used a 70 with 12 blades??? By the way I was hoping to enough airflow to control each side by using 1 servo, landing gear mechanism and the rudder.

Had to hold velocity down so I didn't poke a hole in cover. Joined as a member of club, used facilities when it was not in use and 5 of us flew when the outside conditions were bad.

Look at 30 sec, 1:48 and 1:56 seconds.

 

[telnar1236]

That's why I asked to mount it on one pylon to reduce rotational flow.

Mounting the tail cone on a pylon doesn't seem to make a huge difference - overall it seems better to have the tail cone than not even if you use a pylon to mount it.

I'm not trying to simulate the rotational flow about the motor's long axis in my model or any flow through the motor core since that would add a ton of complexity and require me to simulate the motor mounts as well since they are also designed to recover energy from that rotation on most EDFs. However, even if you could completely cancel out the rotational flow just after the fan, I don't think it would change much as far as my conclusions. Overall, the driving forces behind all of these losses and the direction of flow through the motor are predominantly the pressure gradient along the duct and flow separation from the EDF motor.

 

[telnar1236]

I'm definitely glad you asked about the tail cones though - combined with the tapered duct, it should be very good. Adding the tapered duct to a 90% FSA nozzle has a similar effect to the tail cone and reduces loss back to about 7%.

This also agrees very closely with the values I measured when I was testing the various duct geometries. And adding the tail cone in combination with the tapered duct should put the losses close to zero.

I think with my 80mm EDF with the inrunner motor, the improvements are enough that I want to look into designing a tail cone. With outrunner motors, I need to look into it a bit more, but the pylon is a potentially good option. A pylon-like design for the wires may also help efficiency.

 

[telnar1236]

With all of this analysis, I think I have a good understanding of what I want to do as my next steps. There are just a few questions left to answer, and they're things I need to do experimentally.

1. What material do I want to use? PLA is precise and easy to work with but heavy and a bit brittle. LW-PLA is lighter and very tough due its flexibility but also has a poor strength to weight ratio and is quite weak in tension. ABS is the lightest option for a fast jet, due to its higher strength compared to LW-PLA, almost as tough as LW-PLA when printed with the right settings, and will stand up to the summer heat here in Florida, but warps when printed. If I can use ABS, I want to use it, but the question becomes if it will disrupt the function of the laminar flow airfoil too badly and increase drag. To test this, I'm going to reprint my 64mm sport jet in ABS and test fly it. I like the design enough that I want to not need to worry about damaging it by leaving it in my car, and I have a good baseline. It's also very easy to print and assemble quickly.
2. Do I want to include a tail cone on my EDF unit? CFD says this should be a substantial improvement, so I probably do. However, I need to test this on my test stand to make sure.
3. What does my EDF thrust curve look like? I ideally want to be as sure as possibly my design will actually do what I want before I put my fancy 80mm EDF unit in it. To test this, I'm going to use my other project where I'm printing a bunch of EDFs to donate to my club. I will have a lot of new designs to test fly, so if I stick them in CFD following the same rules I've been using for this project, I can calibrate the CFD and throttle curves to get an idea of how the thrust will fall off with speed. The goal is to produce a plot showing the way EDF thrust falls off correlated to the static efflux speed. This isn't going to be the best correlated value necessarily, but static efflux speed is very easy to calculate when compared to things that might predict the thrust better and is therefore more generally useful. Currently I have 3 points from my 50mm jet trainer, 50mm Super Duper Sabre, and 64mm sport jet, as well as a fixed point at 0,1 which is defined by the measurement process. Flow speed at fan is defined by the speed of the plane multiplied by the ratio of the inlet area to the fan swept area so it's more the part of the speed of the air at the fan that comes from the airplanes speed and not the actual speed of the air through the fan.
   
   

 

[telnar1236]

With how well my 64mm sport jet flew, I also was curious if I could just scale it up to an 80mm size for the final version of this project with the C9 design. Unfortunately, it is a bit too draggy, even with the wing changed to the optimized aspect ratio and wing area from the other concepts for this project and the horizontal stabilizer moved in plane with the wing. My projected top speed with this design was about 145 mph vs. the projected roughly 170 mph for the C7-2. Partially, the inlets are just a bit too big at around 150% FSA on the C9 vs. the 100% FSA inlet on the C7-2.

However, a twin 50mm version of my sport jet, the C10, is a lot more promising. This should have around the same top speed of 145 mph on 4s as the C9 with an 80mm fan on 6s. This means I should have some room to hit my goal of 130 mph on a twin 50mm 4s design which was what I originally hoped to do since 4s is cheaper and more accessible and a twin 50mm design means that I have some redundancy in case one my EDFs starts acting up (not even close to necessary on an RC plane but I've lost a number of planes to EDF failures of various types so it's a minor plus still). The C10 looks quite a bit like a T-38, though with a different tail.

So, I think my plan for this project will be to build both the C7-2 and C10 with the C10 being built first. If the C10 doesn't meet its goals, I haven't spent as much time and effort on it and can try and figure out what went wrong when I build the C7-2 (Super Duper Sabre). And if it works as intended, I will have two amazing planes. I've also continued to work on the Super Duper Sabre design, so this is what it looks like now.

 

[telnar1236]

I finally got the chance to fly my prototype with the rounded inlet lip today and it flew well. The top speed only increased up to 93 mph from 91, but that was without as much of a tail wind so I'm guessing the top speed actually went up by closer 5 mph - not a huge improvement, but still an improvement. It also climbs noticeably better. I think the 50mm EDF struggles with losses after the motor when installed in a duct, so I want to do some testing with a tail cone installed before I finalize my twin 50mm design.

I also did spin testing on my final flight of the day - I wanted to save this till last since I was a bit worried about it being unrecoverable. You can put it into a flat spin without too much trouble - you just have to stall it aggressively and then kick full aileron. Fortunately, you can also recover fairly easily by just neutralizing the stick and cutting power. It also gives you warning before entering the spin - you have to stall it and then keep pushing it through the stall and increasing wing rock until it finally enters it. Overall, I'm ok with this. It's certainly not a trainer, but it's better behaved in a stall than, for example, my F-104 which takes much longer to recover, gives a lot less warning, and is in general less predictable.

With that final test done, I think it's time to retire the 50mm prototype and start building the big airframes.

 

[Houndpup Rc]

I finally got the chance to fly my prototype with the rounded inlet lip today and it flew well. The top speed only increased up to 93 mph from 91, but that was without as much of a tail wind so I'm guessing the top speed actually went up by closer 5 mph - not a huge improvement, but still an improvement. It also climbs noticeably better. I think the 50mm EDF struggles with losses after the motor when installed in a duct, so I want to do some testing with a tail cone installed before I finalize my twin 50mm design.

I also did spin testing on my final flight of the day - I wanted to save this till last since I was a bit worried about it being unrecoverable. You can put it into a flat spin without too much trouble - you just have to stall it aggressively and then kick full aileron. Fortunately, you can also recover fairly easily by just neutralizing the stick and cutting power. It also gives you warning before entering the spin - you have to stall it and then keep pushing it through the stall and increasing wing rock until it finally enters it. Overall, I'm ok with this. It's certainly not a trainer, but it's better behaved in a stall than, for example, my F-104 which takes much longer to recover, gives a lot less warning, and is in general less predictable.

With that final test done, I think it's time to retire the 50mm prototype and start building the big airframes.

Still pretty good for a EDF! 👍

 

[telnar1236]

I got the first few parts of the 80mm version printed. I printed a number of tests in PLA+, ABS, ABS+, Nylon, and CF ABS and I think I'm going to use CF ABS for the first time as the main structural material for my plane. It warps about the same as PLA which is notably better than normal ABS or ABS+ which should retain the aerodynamics I design a bit better, even if my tests with ABS on my sport jet showed that to be less important. I can also print it hotter than ABS without having problems with bridging which means that I can get very good layer adhesion which results in some very tough parts.

The only downside is that carbon fiber requires a ton of caution when sanding since the dust is pretty nasty stuff that you don't want to be breathing, so I've settled on wet sanding (reduces dust) outdoors while wearing a mask and then changing my clothes as soon as I'm done. This might be a bit over the top in terms of precautions, but I'd rather be safe than sorry. I'm planning on painting the plane either white or metallic silver when I'm done with it, so the ugly sanding marks aren't a problem.
One of the big challenges of this plane was the design of the main gear. I've had good luck with 3D printed gear using rubber bands in the past, but it's always been fairly chunky. When I tried to reduce the profile in my F-106 it resulted in all sorts of reliability issues that required a ton of work to fix, so I'm trying a different design from my normal trailing link arrangement in this plane.

This gear strut is much lower profile than my trailing link designs and uses only one rubber band instead of 2-3 which again reduces the size (the thickness of multiple rubber bands adds up surprisingly fast, especially when they're looped around twice). On the ground it seems strong enough, so hopefully it should work well. I'm also using CF ABS for the gear struts since I'm using it for the rest of the plane and it, CF PLA, normal ABS, and GF nylon are all materials I've had success with in the past. If it does end up breaking I may end up switching to CF Nylon, but I'd prefer not to since Nylon is much less accessible to print and requires an additional type of filament.

The wheelbase is fairly wide which should make it resistant to tipping and the nose gear is fairly far back which should allow a very tight turning circle when taxiing. This is useful since many designs can only just about turn around in the space available on the runway on my local field so tight turning designs are much nicer to taxi there. The nose gear will still be a trailing link design since it makes the airplane steer much straighter during the takeoff run. The wheels will be my standard design using GF-PETG hubs (for wear resistance) and TPU-85A tires. The main wheels are 45mm and the nose wheel is 39mm, though I may reduce that to 35mm to gain a bit more space. This plane will certainly not be grass capable.

 

[LitterBug]

YES PLEASE!!! reminds me a Bunch of the F100-C.

 

[telnar1236]

View attachment 254695

YES PLEASE!!! reminds me a Bunch of the F100-C.

I wish I could paint well enough to do a thunderbirds scheme. Currently I'm thinking something more along the lines of one of these schemes but in blue since they're a bit more within my capabilities. Or maybe something like the NASA white and blue paint scheme which isn't a classic F-100 scheme but which I think would still look good on the design.

 

[telnar1236]

Got the wing tips and ailerons printed. You can see the difference between the parts with and without sanding (sanded on the left, as printed on the right).

What may not be clear from the photo is how much smoother the sanded parts are to the touch which should hopefully mean lower drag. Designing these parts was interesting because of the way the thicker back of the laminar flow airfoil and the wing sweep affected the geometry of the ailerons. The aileron is quite thick compared to what you see in most RC plane airfoils.

Combined with the sweep, this means that it runs into the hinges when it deflects. My original design only got about 5 degrees deflection which was nowhere near adequate. Thankfully, I hadn't glued the aileron into the wing yet, so I could remove it and didn't need to reprint the whole wing section.

I rounded the area around the hinges to allow it to deflect up to 20 degrees which is what I originally wanted (the ailerons don't need to be all the effective since they're going to be combined with full flying elevons on the tail surfaces for roll control which is why they're fairly small and don't deflect that far).