Anyone who has kite surfed, or even flown a large training kite, knows first-hand the awe inspiring forces they extract from the wind.  Since my first personal experience with kite surfing in 2002, I've been working on making a kite flying machine for as cheap as possible, as simply as possible, using mechanisms that can scale and be automated, though we deliberately ignoring challenges of automation as these seem orthogonal to a robust foundational design.

I've also kept track of other efforts towards the realization of similar systems.  Some of these seem likely to have also been inspired by the inventors' forays into kite surfing.   Others, found via patent filings, seem to predate the advent of kite surfing.  To get a sense of how wide-spread a notion it is, EnergyKiteSystems.net is a high-recall compilation of links to many dozen, perhaps hundreds of efforts to harness wind energy using kites (along with a variety of somewhat random things).

There are two basic, primary ways to use a kite to harness wind energy.
  1. Use a kite to carry turbines and generators up high and transmit the energy back electrically.
  2. Use a kite as a wing and harness the power transmitted mechanically to the machine that it's attached to.
Both are being pursued actively by numerous groups.

The simpler solution seems to be the latter, if for no other reason than that the part that's flying around is commonly available (traction kites), and the fewer things you need to keep in flight, the simpler.

Once decided on this approach, we need to decide what kind of work the linear force of the kite lines is going to perform.  Again, numerous approaches have been and in some cases are still being pursued, from a laddermill, to a carousel, to other notions.  Trying to keep things as simple as possible, we choose to emulate the proven interaction of kite surfers with large kites.  One way to do this is to have the line(s) wind and unwind spool(s).

Our mechanical-force based approach is basically a long-stroke, reciprocating engine, driven by the tensile force on the kite lines.  We steer the kite in a cyclic trajectory such that the lines reel out as there are large forces on them, and reel back in when there is a smaller force.  As we do this, there's an energy surplus, as the force is greater when the wind is pulling the lines out,
distance_out = distance_in, and
work = force * distance.

Of course, we still need a means of temporarily storing the energy needed to reel the lines back in, and a means to harness the surplus.

Design Challenges

We provide an overview of the physical components of the system by working our way from the kite lines' interaction with the system to the consumption of energy, grouping components by the challenge they're meant to address.

Challenge 1: Steering

First, we have to control the kite.  A typical kiteboard kite has 4 lines: left/right leading/trailing lines.  The lines are attached to a kiteboarding bar so that, relative to the pivot point:
  • The left and right leading kite lines can change relative lengths, decoupled from the net force on the lines (net force is exerted on the pivot point).
  • The left and right trailing kite lines can change their common length relative to the mean length of the leading lines  decoupled from the net force.
If we have a single spool for the lines, we can't steer it in this way.

We describe how to do this in Steering.

Challenge 2: Pivot

Next we consider the motion of the kite in the sky, and how to ensure that the line spools are aligned with the kite lines.  Systems to guide the lines so that they feed to/from stationary spools seem elaborate and likely to place considerable strain on the lines, which we want to be able to scale in size and tension capacity.

In Pivot, we describe how we avoid having to pivot the entire system while maintaining kite line and spool alignment.

Challenge 3:  Drive/Recoil (D/R)

With our solution to challenge 2, our interface is a single line that pulls hard, and then pulls not so hard.  When it's pulling hard, we want to have it impart energy to the system as it pulls out, then when it's pulling not-as-hard we want expend energy to pull it back in.

In Drive/Recoil, we describe how this is done.