I have been busy finding a good way to get the power down from my rotor to the base station. I thought you might enjoy some of my findings - hence I give you:
Open Tensegrity Shafts
Mechanical power transmission using ultralight torsional rigid structures made of components in compression and tension.
Airborne wind energy systems with ground based generators require a mechanical power transmission to the ground. Currently multiple tether-pull based concepts are being investigated (e.g. by the Kite Power Research Group of TU Delft). Pull based concepts require a recuperation phase where the tether is being pulled back in and power is being consumed - not produced.
In today's technology torsion in a driveshafts is the most common method for continuous transfer of mechanical power. Hence looking into torsion based power transfer for Airborne Wind Energy (AWE) seems natural and has been suggested by different sources (e.g. US 6616402 B2 or US8197179 B2 “continuous central driveshaft”). A conventional drive shaft however - even when build with modern materials like carbon composites would likely be prohibitively heavy and cannot be easily retrieved.
Structures made of components in compression (struts) and tension (tethers) as an alternative to a central shaft have been proposed (e.g. US 5040948 A "Driver lines attached to the extremities of the turbines transmit this energy" or US 8197179 B2 “a geometric mesh of interconnected struts” or US4708592 “edgewired turbine”).
This text aims to discuss some of these structures and their usability for Airborne Wind Energy application.
Tensegrity [...], is a structural principle based on the use of isolated components in compression inside a net of continuous tension, in such a way that the compressed members (usually bars or struts) do not touch each other and the prestressed tensioned members (usually cables or tendons) delineate the system spatially.
The term tensegrity was coined by Buckminster Fuller in the 1960s as a portmanteau of "tensional integrity". 
Open Tensegrity will be used in this text for lack of a better term and for a structure that fulfills the definition of Tensegrity except for the "continuous tension" part. An Open Tensegrity structure allows for an external force to be applied to delineate the system spatially. In AWE this force would likely be a lift force generated by an airfoil.
Shaft/Driveshaft can be defined as “a long cylindrical rotating rod for the transmission of motive power in a machine”. For the purpose of this text we will remove the "cylindrical" requirement. We define any rotating three-dimensional long structure for the transmission of power a Shaft.
Open Tensegrity Shaft / OTS Combining the definitions of “Open Tensegrity” and “Shaft” an OTS is defined as a rotating three-dimensional long structure made of components in compression (struts) and tension (tethers) for the torsion based transmission of power.
Open Tensegrity Shafts (OTS)
2-Tether OTS (OTS2)
A set of two tethers spaced by horizontal struts is the simplest possible OTS (Picture 1).
Picture 1: 2-Tether OTS
When twisted the tether force F_T shows two components. A tangential component F_W that will transfer torsion and an axial component F_A that stretches the OTS (Picture 2).
Picture 2: 2-Tether OTS2 - transmitting torsion
This method has been tested and shown to have a strong inherent tendency to hockle/tangle - as can be seen in this video:
When looking at the OTS2 from the top (Picture 3) the reason becomes obvious. A part of the tether force points in radial direction. As soon as the angle between the strut and the center line is not exactly 90 degrees the two combined forces F_R of both tethers create a turning moment causing the strut to “flip”.
Picture 3: 2-Tether OTS2 - Radial force causing taggling
F_R becomes smaller with a smaller angle between the struts which can be achieved by a bigger spacing between them. But this also decreases F_W/F_T and reduces thereby the actual torque that can be transmitted. In practice configurations that seemed sufficiently “flip-proof” could not transfer any meaningful torque at all.
Since in practice catenary sag and wind drag will always bend the OTS which will lead to the struts turning against the center line and ultimately flipping the OTS2 does not seem to be practically useful for application in AWE.
Using ring elements instead of struts has been proposed and successfully implemented by Rod Read . The results look promising and the solution has the advantage of rotational symmetry and less likelihood of tangling. At the same time however the ring elements experience side loads which will likely lead to a higher system weight. You can see the rings in this video:
8-Tether OTS (OTS8)
Adding additional tethers that stabilize the structure can mitigate the tangling problem. The resulting structure (Picture 5) consists of two working tethers that do the actual power transfer (a, b). The working tethers take a shape that closely resembles a double Helix. These are the same lines present in the 2-Tether OTS. Two counter helices prevent the tangling (a’, b’) and four corner lines stabilise and extend the structure (c1-c4). The struts (S1, S2) are positioned in an angle of exactly 90 degree to each other (which makes S1 parallel to S3). The distance between two struts is slightly bigger than their length. This allows them to become parallel to each other when no force is applied - hence allows for flat and easy retrieval of the OTS8.
Picture 5: 8-Tether OTS
The resulting structure has a very high torsional rigidity and can continuously transfer power:
- An Open Tensegrity Shaft is a feasible method for continuous transfer of mechanical power.
- The OTS8 seems to sufficiently mitigate tangling issues of OTSs with fever tethers.
- Efficiency: When using an OTS8 in an AWE system induced drag will lead to low elevation angles of the system and limited efficiency of the power transfer.
- Dynamic behaviour - vibrations and oscillations generated when e.g. by centrifugal forces when spinning an OTS8 around a bend centerline while being dampened by the wind need to be investigated. Since I lack the skills to model and simulate a system of this complexity I will take it to practical testing.
- Dimensioning and Scaling: Finding the optimal strut strength, spacing, rotational speed and tether strenghts while scaling the OTS8 up to useful power levels and finding the lowest possible kg/kW*m.
 Gómez-Jáuregui, V (2010). Tensegrity Structures and their Application to Architecture. Servicio de Publicaciones Universidad de Cantabria, p.19. ISBN 8481025755.
 Swanson, RL (2013). "Biotensegrity: a unifying theory of biological architecture with applications to osteopathic practice, education, and research-a review and analysis". The Journal of the American Osteopathic Association 113 (1): 34–52. PMID 23329804
- 2016-03-10 Added patents US 6616402 B2 and US 5040948 (Thank you Dave and Doug for your input)
Thank you for all the feedback here, here and here. For convenience let me summarize/respond to some of the points:
- Patents: When I quote a patent it is not a comment on whether it is the most relevant on the topic or whether I believe it could or could not be challenged with prior art. It is only to document the state of the art. Many ideas in the AWE space have only been documented in the form of patents - most of them expire before anybody even tries to use them. I am into AWE for the fun, the kWh and the bragging rights - not the dollars. Because patents are patient and the wind is not: When it comes to Airborne Wind Energy there are way too many patents and way to few kWh :) If you file for a patent chances are the only person getting richer is your attorney.
- What about other methods (e.g. loops, belts etc.)? Is OTS is the best power transfer for AWE? I do not know. I do not even know if it will be feasible in real world application at all. Many things can still go wrong - see the "Next steps" paragraph for things that might make me wish I had picked a different method. So far it looks promising and none of these points seem to be a roadblock:
- No large-scale similarity cases (I agree with Rod: "We've never had a great engineering need for torsion tension before. Now we do."
- square-cube mass scaling-limits