I have been running an ablation study for the design space of the foil and its movement.
Varying its size and angle of attack during the sweep in different ways.
The simulations are not yet mechanically coupled, meaning the foil's position is preprogrammed.
Halving the foil chord from 125mm to 62mm seems to buy about 3-4 percent efficiency while tanking thrust about 40 percent in regimes I've been testing against.
But at a leisurely swim it requires 18% less leg power input, mostly because 125mm foil there is not very efficient, if driven like 62mm one it accelerates - 17 Watt vs 14 Watt
Thus the foil hydrodynamic efficiency can be brought to about 86%
The next big design knob: the foil angle of attack, about 20% of the cycle it does almost nothing (less bad for foil half the size, cause it's just smaller relative to kick)
I ran a few experiments for angle of attack: made it mostly maintain an optimal one, derive it from the fin geometry with stops and springs, and added frictions resisting roll - the idea is that right at reversal we lose most of the potential useful work foil can be doing thus if the foil resists that sudden change the friction would be doing useful work for us
And it even works! a little bit of friction adds almost 1 percent efficiency and a huge +25% thrust!
But adding a bit more of it results in flow separation and a we're cooked:
The runs with non-physical angles of attack I tested: cosine AoA and the one where it maintains best AoA for thrust in a square.
Their efficiencies were not that good, cosine was close to passive, square was about 4 percent worse but produced DOUBLE the thrust, cosine produced 1.8x thrust.
I guess the whole hinge engineering comes down to the single question now: how do we make it produce the cosine AoA or remain as close to it as possible for as many operating regimes as we can.
Here's the torque that must be applied to the hinge to obtain that cosine AoA pattern for the current design (I think the shape of data will be similar for other designs as well):
And would you look at that! the acos angle of attack is basically a spring (don't worry about the friction one, spikes are where it hit the angle limits basically)
A single springs can't really serve that ideally everywhere, so the design process needs to continue.
The next step basically is to have coupled CFD - where the physical feedback from springs, etc gets fed back into the simulation.
@7BDiver yeah looks like a torsion bar is an appropriate thing for such a design.
But I will experiment with other mechanical things once I have coupled CFD, another option is to have something that just resists roll - the orange bar above shows that it can resemble the spring quite closely; BTW I got about half the thrust winnings of the spring with it.
I'm pretty sure there are other things we can think of as well - maybe the rubber thingy that dol fin did, IDK how to model it though
