I've been working on the construction of a mechanical prototype of this formation, to try it out in water and see if it would spiral forward. But the 'step and repeat' calculation of each rotator kept bugging me; "what is the factor to make it go round?" Turns out it much easier than I first thought, for each curve (180°), build up out of 8 rotors, this gives the rise to a connector-part with a corner of 22.5°. The same connector can be used everywhere, the only thing that needs to change is, give it a turn at the 4 corners (90° shift of direction). The cool thing is that for the whole composition the rotator shift doesn't play a role due to symmetry. So my guess would be there is at a point a 90° shift, a 'click' happens and the flow can keep on spiraling naturally.