Dream 2700 | A Tailless Tale
Part IV: The Sky Awaits!
Those who have not yet done so may want to read the first three parts of this series before continuing with this article — Ed.
Finally, after a long journey, I was ready for the maiden flight. What a frightening day. I’m not a member at any model airfield right now. More than 30 years ago I was flying in the F3A category, and I was a very active pilot. In the last 20 years, I moved first to paragliding and after that to full-size sailplanes, and my scale model piloting skills went down a lot. I just flew F3K in the last ten years, because it was much simpler to carry this small sailplane with me, without all the hurdles and complications coming from more sophisticated models. My thumbs were shaking and, I must admit, I was tempted not to fly the model at all: after almost two years of effort, I didn’t want to crash it.
Feathers shall raise men even as they do birds, towards heaven; that is by letters written with their quills. — Leonardo da Vinci
At the same time, I wanted to prove my theoretical calculations, and some empirical assumptions were correct. And now I can confirm the quality of the design with pride!
There was no crash on the first flight, and the sailplane showed nice behaviour, without any bad habits.
Getting the center of gravity (CG) right is fundamental, especially on tailless airplanes, where a mistake can easily compromise the maiden flight. A CG too far aft can make the sailplane unstable, and a CG too far forward can make elevon authority vanish. All depends on your calculations or assumptions. I went for a conservative CG at 230mm from the leading edge at the root, corresponding to a 30% static margin.
In a nutshell, the static margin represents the safety margin you have against the boundary situation where the CG lies exactly at the neutral point of the wing. Regarding control deflection angles, the limits were just based on my experience.
Round 1 Test Flights
Take Off
I was worried because of the wing loading (42g/dm²), therefore not so sure a hand launch was going to be okay. I’ve thought about a dolly or even a bungee launch. The dolly was not an option: due to the wing sweep, the CG was falling a bit back after the wing root cord, requiring a specific dolly to be built. At the end I went for a hand launch, being confident that the static thrust (11.5N) was good enough to quickly accelerate the model. First lesson learned: fat fuselages are difficult to handle at full thrust! The risk of a premature release is there, with the fuse slipping out of your hand.
Climb
Just after releasing the Dream 2700, the plane looked very stable and controllable, the climb was extremely good — maybe too much at full throttle. As soon as the plane accelerated, the climb became steeper and steeper, up to the point where the trajectory became almost vertical. Second lesson: thrust line needs to be fixed. The propeller has a thrust angle of five degrees negative, and so the thrust line goes below the CG. I did it on purpose to reduce what I was expecting as a tendency to pitch down. Maybe I used too much. To fix it, there were two options: change the thrust angle — this would be the right way to do it, but it will require a modification of the firewall — or fix it electronically, mixing the throttle with down elevons. This is a bit crazy: you solve one issue, but you generate more! Luckily for me, I went for the first option, adding an inclined spacer on the back of the firewall.
Cruise
After reducing power to 50%, the plane flew very well — like it was on rails. I did not observe any yaw or dutch roll tendencies at any speed. The speed looked good, around 10m/s (estimated), an indication the aerodesign was properly done. Elevons were responsive, only at low speed I see some authority issues . I did not see any pecking or high frequency pitch oscillations. This was observed previously on the SB-13 Arcus full-size sailplane (see Resources). The Dream 2700 has a bigger sweep angle and much higher twist, that I think is making the difference.
Turns
And here is the most significant achievement: turns are coordinated with no tendency for adverse yaw. After roll banking, a small nose up input is needed to keep the altitude. Turn reversal, from +45 to -45 bank angle still needs to be evaluated: the behavior looks a bit slow and ‘softy’, but nothing to be really worried for a non-aerobatic sailplane. Barrel rolls are nice and easy. I cannot say the same for rolls on the axis: the design is simply not made for this.
Yaw stability is high. For sure this is connected to the fins’ surface area. I’m sure I can reduce the fins size and surface, or even fly without them. This is a very interesting test I want to perform in the future. A tailless swept wing with a bell shaped lift distribution should not need fins at all.
Soaring
During the first flight, I forgot to program the engine brake function, and the propeller windmilled all the time during soaring. It’s like having a drag chute attached to the tail! Nevertheless, the glide ratio looks good. I tried to slow down, and the behavior still looked good, with a bit reduced elevon authority, but still with a very steady soaring path. The wingloading is quite high, but the plane shows some capabilities for thermal flight. Applying 6° flaps reduces the speed, but requires some elevon down-trim. To be further investigated is whether the CG position can be further optimized. I’m planning to install a GPS and a pitot tube, to better evaluate the glide performance.
Flaps
The initial setup of the flaps is in three positions: thermal (6°), low speed thermal (12°), and landing (30°).
Third lesson learned: thermal flap deflection needs to be much smaller than that. Even the 6° deflection is too much for efficient thermalling. You get higher lift, true, but you get as well a huge increase in drag.
The design intent was to have neutral flap behavior, as confirmed by XFLR5 simulations. Reality is a bit different, since flap deflection generates a pitch-up tendency. On a second prototype, I want to extend flaps towards the wing tips to correct this.
The first two flap positions are usable, with trim corrections. The landing position was unusable, due to a strong pitch-up tendency: the glider starts raising the nose, the speed goes down, and you easily get close to the stall incidence angle.
Minimum Speed
In a second flight, I tried to slow down as much as I could, acting only on elevons (no flap deflection). At full elevon deflection the plane starts stalling in a gentle way, recovering, and stalling again: not a bad behaviour. It reminds me of what I have read about Burt Rutan canard airplanes’ stall. I didn’t try any bank stalls for the time being. I got the feeling that control authority is not the best at low speeds: this needs to be further investigated. The small chord at the tips, coupled with low speed, produces very low Reynolds Number: this may lead to think about flow separation, but this is not the case. At slow speeds, thanks to the high wing twist, the incidence angle at the tips is very low, and there should be no separation.
Round 2 Test Flights
After the first flights, I implemented the following changes:
- Air intake on the nose and exit on the tail, to improve motor cooling
- Removed 10g from the nose, moving CG to 233mm
- Reduced thrust angle to -1.6° (first flight was -5°)
- Heavily modified flap extension angles
I’m happy to report that all the fine tuning and corrective action produced an improvement of the flight behaviour. With the reduced thrust angle, hand-launch improved, making the plane less prone to nose-up. Cruise at 50% power improved as well, requiring almost no trim. Here are the revised flight parameters:
A nice trial I did: I programmed one of the radio switches with circa 20% engine power. This is almost enough to counterbalance the Dream 2700's overall drag: this means that you can maintain altitude while consuming little battery energy, and even use very light winter thermals to climb a bit!
Flaps are now fully usable, even if I think I will further reduce the flap extension for the two thermal modes, moving from 2.5°/5° to 1.5°/3°. Landing position is now usable, even if it requires some down trim.
What Comes Next?
Despite being happy with the flight results, I’m already thinking about a second prototype with some more design changes:
- A more streamlined fuse, to minimize some of the issues highlighted by the CFD analysis
- Fins with reduced surface area
- Increase the elevon cord, and extension of the elevons up to the wingtips: this should increase control authority at low speed
- Increased wing section thickness at the wing root
Since flying performance is good, I’m now starting to think about making few items available for friends, but this requires a much quicker manufacturing process. I am tempted to go for full molds even if this becomes really expensive. Maybe the best compromise is to manufacture the mold of the fuselage, and continue to use the current vacuum forming on foam cores for the wings. And, whatever I decide to do with the molds, I definitely need to simplify the wing spar and joiner manufacturing process. Any suggestions from the community are more than welcome!
Wrap Up
Somebody once said “the journey is more important than the destination”, and I now know that to be true!
Designing, building and flying your own creature — it’s a transformation journey where you bring your ideas to life. It is as well an intensive learning process. It doesn’t matter if the final product is perfect, what matters is the knowledge you acquire, and the personal satisfaction and pride you feel at each small step.
I would strongly suggest that experience to all of you, and it doesn’t matter how complex your project is. What really matters is what you learn along the journey.
©2023 Domenico Bosco
Resources
- Dream 2700: A Flying Wing Design and Build by the author on YouTube. — The collection of construction and flight videos about the author’s Dream 2700 project.
- Akaflieg Braunschweig SB–13 Arcus from Wikipedia. — “an experimental tailless, single seat, Standard Class glider designed and built in Germany in the early 1990s…”
All tables and images by the author. Read the next article in this issue, return to the previous article in this issue or go to the table of contents. A PDF version of this article, or the entire issue, is available upon request.
