The “backfly” is a maneuver to fly a flexible kite backwards by acting on the trailing edge. On a 4 lines kite we act on the two back lines as “full brakes” to obtain a “backfly”. Quick chained steps are made: releasing, pulling, releasing, pulling, in order to control the fast descent.

The topic Vertical trajectory for yo-yo AWES? mentions this not orthodox way of flight without using the “backfly” term.

Now there is some videos about flyback below:

The video below perfectly shows a backfly, at 8:10, the pilot saying “full brakes”, then a complete cycle upward flight and flyback between 8:30 and 8:42 (note that the speed of the kite is roughly the same during upward flight and flyback while the kite is fully depowered during flyback):

I verified the full depower during the flyback (low quality video as I both piloted and filmed):

The control of a flyback should be possible by automated management, using suitable algorithms and tests.

For AWE it could be a way to realize a full yo-yo mode, the kite flying upward during reel-out power phase, then flying backfly during reel-in depower phase. This sort of trajectory can allow the kite to stay in its lane. So A kite-farm maximizing the space and land use, mitigating entanglement risk, is more possible.
In the sketch below a fictional huge kite (span = tether length) is replacing with several kites side by side.

To borrow an image, let’s say that the space occupied by an AWES farm is like a racetrack or an athletics track, with lanes for each of the participants. If all participants use their respective lanes by using something similar to vertical path, the race can take place. If only one participant uses something like eight-figure with large radius turns he will have to encroach on the neighboring lanes, and the race cannot take place.

I’d love to look at a swaying field of pumping kites.
Much like a massive wheat field wind pattern.
Coordinating the motion of the many kites for their fastest sweep through the sky is still needed here.
To keep them safely separated in a fanned out pattern will require 5 (in the drawing) yoyo control systems in this one ground deployment location.
A single point location deploying to a large ground area has kites flying at the edge of the wind window (weaker) and has less stability as the anchoring is from a point.
Also reeling back on a back stalled kite isn’t fast like the current yoyo recovery procedure of shooting back.

As for a current envisaged kite-farm, with the main advantage of a reduced space compared to the full area swept by the kites.

As I mentioned implementing only one “fictional huge kite” would be a theoretical possibility thanks to no turns, but as the kite span would be roughly the same as the tether length (to maximize the space), a 1 km tether length would lead to a one km span kite, so a too huge kite for now. So said kite is divided into several smaller kites side to side. The whole kite-farm is in fact a fractal structure.

It would be the same for a single huge kite of 1 km with a one km tether length. It is the small price to pay to maximize the space. And also we are do not have to occupy the entire flight window.

To figure the set a kite-farm would be like a 10 m span 10 m lines paraglider but in higher scale, leading to the division of the paraglider into several identical unities side to side, as for a fractal structure.

With a one km tether length a flight window of 120 degrees (some kites can fly well enough at the edges) would be 1046 m wide, let’s put 1000 m. Assuming an altitude between 200 m and 600 m. So the swept area is 400 000 m², and is worked in crosswind flight, by using only about 4 km² on the land by taking account of all wind directions. The Power Harvesting Factor ζ of a flexible kite with a glide (L/D) ratio of 5 would be 4. So 100 000 m², or 50 000 m² to be conservative, of kite area could be implemented in such a space, compared to only a few 1000 m² for envisaged currently kites flying with large eight-figure eating all the space while harnessing little due to large radius turns, so about 10 times more.

Almost all curves show that the reel-in phase (recovery procedure) takes about half of the full time.
As the videos (see the second video between 8:30 and 8:42) and my tests showed, the speed is roughly the same for both upward flight and backfly, so the time of the recovery phase would not be affected.

Moreover the pull is lesser (almost zero) during backfly while the consumed energy is about 1/4 during current recovery phase.

Some minor difficulties have to be overcome such like the automated control in order to avoid both loss of time during backfly and crash risk (which is mitigated as the lower altitude can be limited to 200 m as I mention above), even by considering that a crash on the trailing edge would not be too serious, and could be seen also as a landing when it is controlled.

Indeed the backup also allows to facilitate landing and takeoff operations.

As a temporary conclusion, using backfly in a kite-farm configuration such like I detail above, using vertical path, would lead to the next order of magnitude of the power/space ratio. So the control of the backfly can be a major step for AWE.

Kites are very difficult to control in “backfly” mode where they may be mostly stalled and providing only drag. In windy conditions, the return to normal flight will be very “explosive”.

I doubt this is a good way to go. A wing is better used in the region of C_L zero to its maximum value, but not increasing angle of attack further to introduce stalling.


1 Like

Alternating “backfly” and upward maneuvers are perfectly controlled manually (where is the “explosion” on the video between 8:30 and 8:42? And also by using a single skin wing, inertia would be mitigated). So It should be the same with computerized means.

Computerized tools don’t still achieve takeoff and landing of a flexible kite unlike for manual control. But the difference is that takeoff and landing are tested for a while without any success. “Backfly” mode has not be still tested with computerized tools, even as means such like flaps destroying the lift are studied.

However I don’t have an explain for how the kite flies “backfly” fast (roughly as fast as upward flight) without generating pull. I noted this in my tests. And we know that we make no power during “backfly”. So this mode should be analysed deeper for yo-yo reel-in phase as it would allow to achieve a far better management of the space/use used with vertical paths (that works as lanes).

We can of course immediately dismiss this idea or other ideas to try to improve the power/land ratio, continue on the same path, and repeat the same failures (in addition to normal engineering issues, no work on the management of the space, so no market) from conference to conference.

See Fig. 23.13

Anyways it doesnt matter. Either my comment is good or not. I dont think if I am right that it matters that I am working against ideas to improve land usage.

I just think a wing flying in stall or backwards is not a good idea. We should rather come up with some other ideas, or even better prove me wrong.

There are many better ideas out there not discussed. I know because I have come up with a few myself, but I am not at liberty to share. Also from looking at KPS stuff they did think of a few novel things. I am not at liberty to share that either. But there are still many paths not followed.

That being said, with increased understanding of the possibility of AWE, the problems seem to appear as fast as new opportunities. I may end up like @dougselsam, thinking that there are very few options left in AWE (sorry for «quoting» you on this it is maybe even not near correct). But I am not there quite yet.

I will try to be more constructive in some feedback:

One issue with Yoyo at large scale is that most generators on ground can’t really change from reel out to reel in in an instant. They may need many seconds to change speed. This is one reason why many yoyo outfits (like Kitemill) intend to do several loops of production and then a single retun phase, rather than doing production and return in one and the same loop.

The energy generation cycle you describe sound like having quite frequent changes between in and out, and no way to do many loops in sequence without return phase.

Also a kite that is «backstalled» (just stalled really) will probably pull a lot more than a kite at C_L = 0 (pointing leading edge directly into wind). If the wind is, say 8 m/s and you want to reel in at 16 m/s, the drag in backstall wil increase by a factor of 9 (due to dependence on drag to relative airspeed squared). So you may create a power plant with net zero production? Energy is given by force times distance travelled.

Also, a kite that is motionless in the wind window probably doesnt have ideal flow conditions, unless it’s a rigid kite with a single attachment point. From the time the kite is still until its moving at full speed, it will not generate power efficiently, and thus will shave time off the valuable production cycle.

So finally to reiterate my initial point, backstalling may be fine for low wind conditions but I dont think it scales so well to high wind occasions. I can explain this by saying that a stalled kite has turbulence and is erratic, and suddenly may accellerate forward. In this way it will get proper flow easily through apparent windspeed. If this is the case, the plant will have a small capacity factor due to high wind nonproduction time.

Most of these problems are solved once you let the rig rotate. Now you only have to deal with differen speeds at different radii.


On the above I see many opinions and hypotheses, but specifically no (even manual) tests or deeper analysis to support them, although they may be admissible, the backfly being not an orthodox way of flight. That said backfly is used in kite field.

Let’s take a look at it from the factual to the questionable. The backfly flight is fast as shown in the videos. It is a fact. Furthermore I could contend that the traction was almost zero. But I would say it is a feeling more than a fact. However, I am not aware that the backfly is used to generate traction. From there the question arises, how does this happen? What are the forces at play? Gravity? Maybe a negative lift? These are questions that only further tests and theoretical studies can answer.

A subsidiary question can be also: is the control secure enough? Suppose this is not the case. The other subsidiary question could be: what are the risks? Is a fall of the deflated wing on its trailing edge destructive? Would the possible loss of time during reboots destroy the power resulting from the overall density of the installation?

Then we would have to compare with what exists. To begin with, we must consider the inevitable time losses for the yoyo mode, as well as the losses due to traction, which are known and measured (see pdf above). Furthermore, what are the risks of crashing from VTOL mode to flight mode and vice versa? Are these risks increased due to the higher cut-in speed with a rigid wing? What are the consequences of a crash with a rigid wing?

Now the problem being the management of space and one of the possibilities remaining the vertical trajectory, the question can be: is there another possibility of making these back and forth always without turn (turns take space) but without backfly? The answer is perhaps, if the wing is reversible, able to fly from both sides, and moreover probably being limited as to its L / D ratio. Let’s take things further: can we manage space without vertical trajectories? Probably yes, with horizontal trajectories to start with, or perhaps much tighter figure-eights or Low radius loop.

There are also of course the rotary devices to be able to better manage the space, but they have another problem: they do not seem to be able to exceed a few meters of altitude, and this for years. I would add that if we used the kite lifter alone in crosswind yoyo mode, it would produce more energy than the rotary assembly.

Besides this, I don’t really believe in decisive ideas that would be hidden for years. When something works, we want to show it.

Please can you support this statement by providing some reference? Thanks.

No, as it is a vertical path, the only relative airspeed coming from the vertical descent of the kite, not from the wind speed, even less the wind speed added to a relative wind which does not occur against the wind direction. Indeed reel-in speed of 16 m/s can occur while the kite flies upward then “backflies” vertically without approaching or moving away from the station.

An example of vertical path is given on the figure 1.18 on:
The trajectory is almost vertical. The reel-out and reel-in distance is deduced from the elevation angle of the AWES between the bottom and the top of the trajectory. As a Magnus-Effect-Based AWES would produce energy by using this vertical trajectory, one can deduce the same for a “classical” kite.

Au Contrare Mon Frere I have been saying for a decade “nobody even tries the simplest things!” To me there are too many possibilities to count, Nobody bothering to try even the simplest ones, most not requiring much of a brain or much of a budget or much of a computer other than to shop for parts, meanwhile what I see being tried seems unlikely to prevail, (happy to be proven wrong on that).
I mean, what, a guy just flew a kite crosswind to directly spin a generator using a rope for the first time in 2018? After a decade of people repeating the “crosswind kite power” mantra like a herd of zombies? I told everyone a decade age, when you talk about wind energy, if you have to say “crosswind” you probably don’t get it. It;s how it;s been done for a couple thousand years. Heck a hundred college kids should have tried that by now. I could think of probably a hundred possibilities to try, actually more than that. Way more. Good ones too. Meanwhile we have “a…kite…can…pull…a…string”. ( I wanna say. what are we in Cub Scouts?) Well, OK, good luck with that, and I hope it does work. :slight_smile:

Drag coefficient of a plate around 2. Of a streamlined wing 0.05. Which mode would you choose to pull the kite in the return phase? The area of the kite determines the lift, so it appears the «backstalled» kite would have more drag (much more) compared to the wing at zero C_L.

1 Like

I see. You foresee more of a drag mode setup. That indeed changes things a bit. I would guess that would require a second tether from the kite that is more (most?) vertically inclined connected to a second winch?

I dont have much faith in systems that are not mostly crosswind. The test for this would be that the kite is flying faster than the windspeed. There are many reasons for this that we have discussed numerous times before in this forum.

  • low use of kite materials
  • low maximum theoretical extracted energy
  • litte in terms of handling high wind speeds

I would say the same, your idea has not been sufficiently detailed explained. If it had, I could have shown you where my doubts were.

If you are ok with discussing half-baked explanations (as I am), you should not expect scientific quality objections in response. Still, I could easily produce such equations based on a long tradition of aerodynamic studies. Flying a kite backwards is unorthodox but still may be easily explained in sufficient detail but aerodynamic theory.

To give some input on what may improve my answers:

  • better description of the kite flying path you propose
  • what lift and drag you expect along that path
  • how you will produce energy
  • tether length and diameter, number of tethers
  • kite area and number of kites
  • kite design (rigid, inflated, C kite, single skibs …)
  • number of kites

Anyways I think its not necessary, I should already have provided enough input to scrap the idea in its initial form (a yoyo awe rig using backstall for return mode is what I thought that would be) and see if there are variations that hold more potential.

Strange you say this but mostly provide feedback that everyone else doesnt understand wind, but never provide any ideas tangible enough for the rest of us to comment. (the Sky Serpent is a good counterexample, but hardly counts towards a hundred possibilities to try)

1 Like

You quote some generalities about some drag coefficients, without replying about “will probably pull a lot more than a kite at Cl = 0”. Have you a reference of the pull during “backfly” compared to Cl = 0?

I don’t think the shapes in the table apply to the shape of the kite during “backfly”, which looks to be like
a flag without real consistency flying by the trailing edge (see the video between 8:30 and 8:42).

During my tests with a 4 line 4 m² ram kite, I was literally pushed back during normal upward flight while I felt no traction during backfly. But as I have already indicated, these are only impressions to be deepened.

No, there is always a single tether and a single winch, as for the Fig. 1.18.

I’ve had time for a lot of walking with the 4 line Peter Lynn skin kite. It can fly backwards.
It pulls super hard when looping as per normal kites. It can perform super tight loops in light wind.
Flying it backwards is tricky as you have to go through stall and the new shape is kinda horrible. It can be flown up and round backwards. Not as easily as say a rev kite of course.
Stacking nets of revs could get @PierreB the upwards pull we want. A backwards and reeling in flight with a rev would be hellish. Probably better to pitch the whole set to dive forward.

@PierreB you mentioned being disappointed at progress in scaling rotary designs. Me too. I wish I had resources and support to do just that. I get what to me seems like sweet F A. Don’t assume this lack of progress in any way reflects the amazing results we get from kite turbines.

Crosswind mode are not only figure-eight or loop. A vertical path works also in crosswind mode, the kite flying several times the wind speed in a similar way. Test a kite flying in the center of the flight window and you will notice it as I did by being pulling back strongly. On the video it is clear that the kite flies far faster than the windspeed.