Steam Resonator design session v1: reasoning

Base grid sweep, 20,736 candidates, all four geometry classes crossed against the 1200V / 10 kHz sequential-switchover drive with the 10Ω pickup plus 500Ω resistive choke doing the amp restriction. The top five come out in a dead heat at score 0.386, ~0.62 kg/hr steam, and they span wildly different cavities — a 3.52mm gap with 137 mL and a 1.5mm gap with 24.5 mL score identically. That flatness is itself the finding: the score is tracking delivered watts, not any resonant feature of the geometry. Anything that dumps ~450 W into the water boils about the same amount of water.

And that is the ugly part I have to say plainly. Look at the heat budget on the winner: dielectric relaxation contributes 2.98e-5 W. The double-layer cycling and ohmic leakage together contribute 448.8 W. That is 0.0% non-ohmic — to nine-ish decimal places, none of the cavity heat is coming from Stan's "particle impact" dielectric mechanism. This design boils water by pushing 643 mA of leakage current through it and warming it resistively. That is a kettle with a fancy choke, not a Steam Resonator. The resistive pickup and choke are supposed to restrict amp flow so the field does the work; instead the leakage is the entire heat source and the dielectric channel is thirteen orders of magnitude below it.

Why: we are running at 10 kHz, and water's Debye relaxation peaks at 56.9 GHz. We sit at 1.76e-5% of the loss peak — ε'' is essentially zero at this frequency, so there is no dielectric heating to be had here by construction. Worse, τ_bulk is ~10,000 ns and the on-time is comparable, so the field only just establishes before the phase flips; the bulk field is real but it does no lossy work at this frequency.

Next session: stop rewarding delivered watts. Add a hard penalty on ohmic fraction so the optimizer can't win by conducting. Then sweep frequency up by decades — 1, 10, 100 MHz — to climb ε'' off the floor, and re-examine whether the sub-critical Townsend film, not bulk dielectric loss, is the real non-ohmic channel. Log the null result; it defines what to change.