From an aerodynamic point of view, I have to admit that I haven’t looked into the matter. That said, planes can scale up to limits that we don’t perhaps know about, and in any case beyond a wingspan of 100 m, and wind turbine rotors up to a diameter of 260 m and more.
But perhaps I didn’t quite understand the question.
From a structural point of view, this remains to be verified, but transmission by external motorized belts on an inflatable flexible cylinder should help considerably.
Now there are some other points to verify, whose some concerning the vertical trajectory which was theorized but not tested (unlike almost elements of Omnidea’s balloon which used basic oblique and downwind trajectory). The chapter 13 (second AWE book), pages 313-315, mentions a lot of elements to consider for a more accurate calculation of the expected efficiency, and summarize the advantages of the vertical trajectory:
The results of the optimization, providing average power of 89.18 kW, are significantly better than the basic control strategy results, which produced average power of 40.15 kW. This means that an improvement of 122% has been achieved through optimal control. Note that the basic control strategy causes radial lift, in contrast with the vertical lift associated with the crosswind motion of the optimally controlled system.
The advantage of the vertical trajectory seems obvious, but the apparent wind could also deviate a bit the “vertical lift”.
Other points: a balloon of 1 km long and 200 m diameter has a volume of 31 000 000 m³, leading to an air mass of more than 30 000 tons. I don’t think it is a problem when using the pumping (yo-yo) mode: the rotation is stopped far before reaching the top of the trajectory, and is starting again far before reaching the bottom of the trajectory. Perhaps, during descent, a reversed rotation could mitigate the required force for the recovery reel-in phase. But such a mass can lead to a safety issue: valves are required in order to quickly deflate the balloon if it escapes or if the traction is too high.
Now if such dimensions are not practically feasible, a farm of “small” unities of 200 m long and 40 m diameter would be a possibility. The trajectories would then be staggered. The density would remain high because possible collisions between unities would not lead to serious consequences, as for AWES farm in bumper car mode. And testing such a balloon in plastique film would be possible during a good day, and not too expensive: taking a max of information (power consumption, wind speed, average power…, then dispose of or recycle the balloon if it is out of order after tests. It can be easier than deeper simulations.
On the paper, I have favored experimental measurements wherever possible.
Balloon 1 km long. Ground station 1 km long. Both are parallel. What conflict?
As the paper indicates (sketch page 4) a circular track (7) retains the ground station (6) in order to mitigate the cantilever effect. Some concentric circular tracks could be added.