Kite propulsion has shown some serious potential as an awesome way to power boats while saving fuel and boosting performance. Yet there are a few challenges standing in the way of kite-powered adventures. In this article, we’ll dive into the modern problems faced by upwind kite boats and explore some cool solutions that aim to overcome these hurdles.
Current systems:
Current automated kite systems, such as SkySails Marine, are only designed for downwind use. They have just two actions where the kite’s vector needs to be known:
- 1. To stay neutral and level at the apex of the wind window
- 2. To know generally where the kite is when on a very long line with the kite far above the water (only low positional precision required), doing loops and figure 8s with a relatively slow, low-aspect kite
They reduce these systems down to low complexity and computation, because it simplifies things. That being said, I am very impressed with SkySails’ technology for launching and landing huge low-aspect kites. Much of SkySails’ approach has informed my own. But, as I understand it, they cannot accurately and precisely determine the kite’s vector at any possible location within the wind window, nor when the kite is doing certain sorts of movement or lack thereof.
The K2 Kiteboat is the most impressive upwind kite sailing system so far, but it will be exciting to see the new kite boats in development to beat speed records. The K2’s top recorded speed is 37 knots. It is beautiful and the control system is amazing, especially in how it moves with the kite angle to reduce the necessary righting moment. I think they say they launched as large as a 60 sqm kite on a light wind day, which is awesome from such a small boat. All that being said, their system, like the other existing upwind systems, requires active human control of the kite.
Upwind kite propulsion: Current problems
Launching and landing large, high-aspect kites
The type of kite necessary for upwind use is very different from a kite designed to fly loops downwind. The downwind kite, using a very long line, can generate 9-15x the power of equivalent sail area on the deck of the ship. It manages such incredible increases in power by creating its own apparent wind as it loops around, and accesses the higher wind speeds at higher altitudes.
Kites have more restraints if they are meant to perform when the ship is going upwind. Primarily, they are not able to take advantage of the long line and high altitude winds, and the high kite speeds that are created in loops. So they need to have more sail area, and it needs to be high-aspect (wide left-to-right and narrow front-to-back). This is because, upwind, any propulsion the kite provides is going to be based off the ship resisting the horizontal portion of the kite’s pulling force, so that the ship can be pushed forward by the forward component of the kites pulling force. This means that propulsion benefits from having a large forward component to the force coming from the kite, which happens when the kite is located far forward in the wind window. Downwind forces result from being very far back in the middle of the wind window.
This means any factor that drags the kite farther back in the wind window reduces the boat’s upwind performance in both speed and pointing angle. For example: Any movement of the kite to generate more kite speed and power will result in the kite falling back somewhat within the wind window. A longer line will result in the kite falling back within the wind window, because the longer line has more drag, even when the kite is still. The kite force acting on the boat is a vector in the direction of the kite line as it leaves the boat. This means that line sag from wind drag results in an even lesser pointing angle, not just from how far it drags the kite back in the window. A low-aspect kite, which is used by current designs for their stability while launching and landing, has higher induced drag which cripples a kite that needs to fly far forward in the wind window.
For a good upwind sailing angle the kite has to stay still, to be far forward in the wind window, and generates less force than when it is moving. Therefore, upwind kites need to be significantly larger than downwind kites to generate meaningful power. They need to be about the size of the sails they are replacing, and as described in my Kites vs Sails post, a kite boat can use much more sail area than a normal sail boat without capsizing or pitchpoling. Combined with the fact that high-aspect kites are less stable than equivalent size low-aspect kites, gigantic high-aspect kites present a difficult beast to control, especially when launching and landing the kites. A huge landing area (larger than the boat) is required unless you figure out a system like SkySails yet are able to launch significantly larger and higher-aspect kites than their current system. A gigantic flapping high-aspect kite is difficult if not impossible to manhandle in anything more than a light breeze. Even then it is hard if the wind is constantly shifting, which is common in light breezes.
Reefing
Traditional sails can be reefed to reduce their sail area when the wind gets stronger. So far, no-one has come up with a solution for reefing kites that does not hinder the flying characteristics, such as lift/drag ratio or controllability. Without the ability to reduce sail area, normal sailboats with enough sail area for light wind propulsion could be dangerously over-powered once the wind picked up. Because kites do not produce the torque that sails do, and because kites can be flown at their apex to be pulling the boat up rather than forward, and because kites can easily pull a safety line and fall out of the sky; being overpowered is somewhat less dangerous, that is, until we want to land the kite. Good luck reeling in a giant overpowered kite without a monster winch. Even if we get past that issue, it is unwieldy and dangerous for a high-aspect, sail-sized kite to be handled on deck when the wind is strong and gusty. If we gain control of a wingtip, the kite will still violently whip around, even if the safety system has been pulled. It sounds like a recipe to damage equipment and crew.
How can we solve these issues?
Build a kite robot able to sense its position and direction precisely and accurately throughout the wind window, no matter how the kite is moving.
The most important extension of this is the ability to fly multiple smaller kites instead of just one large kite which is dangerous to launch/land. This enables us to get enough kite-sail area in the air to take advantage of kites’ performance characteristics. The pictured 15M boat could launch 20% more kite sail area than its equivalent Harryproa, while being lighter. In addition, by landing or launching these smaller, manageable kites one at a time, the system can be effectively reefed without reducing efficiency or control.
The idea is too obvious!
This was the doubt I had when I first came upon the many-kites idea. I’m sure others have thought of it too. This system requires a level of precision beyond what current systems are capable of, as the kites need to perform delicate choreography to avoid catastrophically contacting each other. I don’t think anyone else has given this many-kites idea much energy because it requires vectorial precision along with automated coordination that even a team of pro kiters would have a hard time maintaining — they would need to read each other’s minds. I think most people are stuck contemplating how to handle giant kites on boats, and didn’t chase after automated precision because it seems a little unnecessary when the sky is so big. Sort of a chicken and the egg situation. Only a computer can coordinate the kites well enough to fly in such close proximity for extended periods of time.
It was only during a conversation with a friend that I described my frustration with a statement like this,
“It is a shame I can’t get past the giant kite launching and landing problem. If I did, the patterns and performance this system could pull off would be amazing… I swear, it’d be precise enough that we could even fly multiple kites without hitting each other! … Huh.”
A bit of over-confidence in that statement, but I feel grateful for that conversation. Our brain actually lowers in activity when we are speaking. Knowing that, it’s amazing that many of our best ideas come from conversation. It’s as if speaking frees our minds from crippling overactivity.
About the robot.
The robot is to be positioned just below the kite for precise control of of steering line tensions, no matter how far the kite is from its attachment point. Two traction lines from the kite merge just underneath the robot into a single traction line which runs from the robot to the deck, which allows easy reeling in and out of the kite with a winch on deck. The single traction line also means that we can easily move the kite bases around (as long as the kite is de-powered) with a low friction ring or pulley on the traction line. This allows us to stuff as many kites as fit in the sky onto a boat, and move kite bases as needed to launch or land a kite. The winches keep track of line tension, and automatically reel the kites in if the line tensions wither in light wind.
A specific configuration of sensors is required to measure both the kite’s position and direction within the wind window to this level of precision and accuracy:
- An accurate apparent wind direction is calculated from a sensor on the boat, preferably one raised up on a short mast.
- Another absolute orientation sensor is placed underneath the kite pod, on a short hollow rod connected with a U-joint to the kite pod, through which runs the main traction line. This measures the traction line angle as it leaves the kite pod, more important than the angle of the pod or kite itself, and also measures the direction the kite pod is facing within the wind window.
GPS is to be considered for back up tracking, especially at very long line lengths. A tension sensor just above the U-joint would be ideal, as tension determines kite control sensitivity. But I am designing so the chassis does not to take the brunt of the tension loads; so this tension sensor on the pod becomes difficult to implement. I also am unaware of a tension sensor that meets the low weight requirements for the pod. Perhaps a combination of an apparent wind sensor on the robot with a tension sensor at the base winch is enough to predict control sensitivity.
Math
To calculate the position and direction of the kite within the framework of the wind window, we need 3 equations. Looking through research papers, I have not found the specific equations I need, as they were all for downwind power generation systems and were tracking the kite in a different way.
Though I may need a mathematician to help me elucidate these, I believe the equations should not need variables for line length or any other linear variables. They should only need radial variables for the Euler angles or quaternions of the kite and its traction line, outputted by the absolute orientation sensors, and the apparent wind angle outputted from a sensor on the boat.
1. Equate the longitudinal angle φ of the kite line with respect to the apparent wind angle yw.
2. Equate the latitudinal angle β of the kite line with respect to the horizon.
3. Equate the kite directional angle X within the plane tangential to the wind window.
Downwind courses
On broad reaches, there will be a certain downwind angle past which it is more effective to have a single kite on a long line doing figure eights. Before and well past that angle, the boat will perform extremely well with all the kites flying and making turns up and down on 30M lines. I am curious to see what that angle ends up being for fast boats. A high-aspect foil kite looping on a long line will generate even more power in loops, compared to current systems of equivalent kite area, due to the higher speeds it will attain with less drag. Perhaps it is possible to leave the other kites flying in neutral positions for the aforementioned downwind course.
On a light, easily driven boat, such as the 50ft concept boat in the top image, heading deeply downwind, in light wind, with a 22sqm high aspect kite looping around in figure-eights, it would be easy for the kite to accelerate the boat so much that it outruns the kite. This can be a catastrophic failure situation for a multi-line system controlling the steering lines from the boat. The kite could tumble out of the sky, getting wrapped up in its 4 lines. But our pod robot just becomes a tiny little paraglider pilot, and turns to face the wind window until it reestablishes traction line tension. A potential solution is to drag a small drogue behind the boat. Though, I think a better decision is to simply avoid straight-downwind points of sail. I don’t know that outrunning the kite will even be a problem if we stick to broad reaches for downwind routes.
System Energy Management
I don’t want to launch and land kites all the time to change the battery in each robot. Certainly, we can get kite control to not take much power, especially if we are on long tacks with the kites mostly still. But I’d like to fly the kites indefinitely, and so I have a couple of ideas for power. I don’t think the little pods have enough surface area to make solar viable. When the kites are flying on short lines between 10 and 30 M, I think a small wire within the traction line can trickle charge the robot from the boat. I’m debating whether converting to AC on the boat and then converting back to DC in the pod is worth it.
When a single kite is let out on a long line to loop downwind, the wire in the line gets too heavy. Along with weight from extra line length, we actually have to use a thicker wire to prevent voltage drop-off over increasing distances. So I think we need to disconnect that wire at 30m and figure out a different way to keep it charged. On a long line doing loops, the kite speed will be very high, generating lots of apparent wind. So I propose a small wind generator on the robots that get used on a long line. It does not need to be large as it only needs to keep up with the discharge, which, even switching between right-hand loops and left-hand loops, will not be that high. These generators need to have low-drag dormant configurations, as we want as little drag as possible on close reaches.
How does this open up kite potential on cruise ships?
With moveable kite bases, many kites can share the same collapsible landing/launch pad, thereby not taking up too much deck space. On a cruise ship, the landing area could be mostly netting. They could easily suspend it with extendable beams over the deck, leaving the space underneath still usable. Then they could easily retract them back to a central point, to open the sky back up again for the on-deck experience. Passengers could observe the launching/landing process closeup (perhaps directly underneath) while staying safe. 3 to 4 landing areas on the largest cruise ships could service at least 51 kites, 30sqm each. What a stunning visual experience. Imagine the light shows!
Conclusion
The reason there are no kite-powered cruiser yachts comes down to the launching/landing problem and the reefing problem. If we truly can solve these issues, the manifold benefits of kite propulsion are unlocked, as described in my other foundational post. I certainly need to acknowledge that there is much I don’t know that I don’t know. I will keep you updated on my progress, and I appreciate all questions, feedback and support.
Talk to Me
Have questions? I am happy to consider your ideas.
Eager to support the project? You are more important than your money. But if you want to. . .
Hi Seth,
Great website. I have a couple of questions:
How would multiple kites be flown without tangling? It seems flying in a single 0 or 8 pattern would tangle things up and you would need to precisely reverse the pattern to untangle (which is doable). Spatially separated patterns would work, correct? Specifically, how would you do 8’s or ovals at the edge of the window for going upwind with 5 kites?
Why not control all the kites from a single multi-drum device?
Since you don’t require much righting moment when flying a kite, what is the advantage of a proa? Would a catamaran or trimaran not also work?
Thanks for your comment here Klaus!
Doing figure 8s or loops with the kites will only be possible downwind. The kite force vector at the boat is always in the direction of the kite line, and any movement of the kite, especially loops or figure 8s, are going to bring the kite further back in the wind window. The only movements of the kite, that still might be able to get a mild upwind sailing angle, are the kites turning back and forth so that they travel up and down in the wind window, but are still always facing in the same average direction. This can be done with multiple kites as long as their line lengths are similar enough that they can travel up and down at a similar radial speed (longer line lengths mean the kite has to travel faster for an equal radial speed). They must all make the turns in synch to avoid hitting each other. This maneuver will be useful from a beam reach to a broad reach, and, again, might be able to go slightly upwind.
For a close reach upwind, the kites need to stay as far forward in the wind window as possible, necessitating more kite area and minimal drag. This means high aspect kites, with short lines (15m-30m), staying still. In this configuration, they have basically the same apparent wind performance that high performance wingsails do, with minor aerodynamic benefits. But, kite boats need less righting moment, as you mentioned, so they can have less wetted surface area and weight, and more kite ( sail) area for increased power.
Next, when going downwind, looping or figure 8-ing on long lines (100m+), one kite can potentially. produce the power of 10, so we don’t need more than 1 or 2 kites flying. At those long line lengths, there is plenty of room in the sky to do loops or figure 8s with two kites without needing to tangle the lines.
If the single multi-drum device is what you were describing on the Harryproa forum, my main argument against it is that each kite (except the top one) needs to be flown at an angle to the main line, meaning only a part of the tension it produces is transferred to the tension felt by the boat. I feel that this reduces the drag benefit of the single line in that system, as we now need more kite area flying for similar power felt at the boat. As I have said before, I also think it complicates the location sensing, but I am looking forward to seeing how your solutions address that challenge.
Finally, theoretically, a planing non-proa boat can have the least wetted surface area and therefore the highest performance with a kite. But, besides the righting moment potential for sails, proas have many benefits. I personally want more livable surface area on my boat than a super narrow, short monohull, and shunting is much simpler than tacking or gybing, especially/even with kites. I also like that shunting lets me easily back up, and it is what I am used to doing on a twin tip kiteboard. There are other benefits described here on the Harryproa site, and basically anything there that does not have to do with righting moment, is still very attractive to me on a kite boat. Plus, Harryproas are simply safer than other sailboats.
Hi Seth,
Thanks for the detailed answer about how best to fly multiple kites in various scenarios (upwind, downwind, reaching). I agree when going upwind that one needs to park the kite near the edge of the window and that line drag, i.e longer lines, likely reduce your pointing ability, unless they allow you to fly the kite higher in less turbulent and stronger wind. Most kitefoilers use 4 lines to control the kite and keep line lengths to 12m + 3m bridal. You are proposing a pod with one line (and a depower line?) reaching down to the boat, hence this will reduce line drag a bit. But for such short lines is the added complexity of the pod really worth it? Also, with short lines the kite will react with much larger angular changes to gusts and turbulence. Do you think a pod will be able to handle this?
Regarding line handling and controlling kites on the boat, distributing multiple winches across the boat may help you fly multiple kites on short lines as this would space out the kites and their lines a bit. But would this be sufficient for sailing upwind where all kites are crowded into the lower edge of the wind window. With short (15m-30m) lines it seems difficult to not have kites aerodynamically interfere (act like biplanes) and tangle their lines. Possibly having different line lengths for different kites may work. And maybe this can be tested with multiple kitefoilers going upwind in close proximity?
I agree with you that Harryproa’s are a great design with lots of usable space. Indeed, shunting comes naturally when you have done some kiteboarding before and I agree that shunting has it’s advantages. Nevertheless, the complexity and weight of bidirectional bows on a Harryproa vs. the simplicity of steering and auxiliary propulsion on a unidirectional boat would make a cat or trimaran an attractive boat as well. I do however dream of hybrid approaches, i.e. having a Harryproa with telescoping masts with some additional kite propulsion …
Cheers,
-Klaus
To clarify, the 2nd line for depower is only intended for the initial prototype, for less complexity in activating the speed system on the paraglider. It is not my first choice for the performant production version I currently have in my head, which would use a kite similar to the flare moustache. My current idea for that is to use rear line depower by having two spools of line on a differential gearbox.
I am assuming it will handle it. Even for short lines, I like the pod because it allows the single line, which I imagine is easier to move around when positioning the ‘kite bases’ for launching and landing. It also provides a better place to mount orientation sensors, for the reasons I discussed on the harryproa blog.
My idea for how to launch and land kites is pretty mundane. It is just like launching one on the beach, on a very short line. I should make some 3D mockups to illustrate. Getting the kite line base into the right position is essential for this in my mind.
I am keeping my fingers crossed that we can have 5 kites flying in a configuration that isn’t hindered too much by aerodynamic interference. This will likely require different line lengths, but I don’t think those line lengths differences can be too great as longer lines will hang back farther in the wind window and not perform maneuvers as quickly as shorter lines, so it may be difficult to coordinate the kites during tacks, gybes or shunts.
I would have at least one small/telescoping mast on my kite boat in a place where it doesn’t interfere with the kites. Easier to do this on a proa, I think. The mast would be for instruments and backup propulsion with a small sail.
Also, the GPS stuff you touched on in the blog is very interesting, and a path I need to consider more. I’ll start with the orientations sensors though.