EPG design session v5: reasoning

EPG sweep, 24 Aug 2026 — 480 candidates. The intent this session was threefold: break the I²R wall by dropping to heavier gauge at fixed 2400 turns, push flow velocity past the laminar knee to see whether induced EMF scales with velocity faster than hydraulic drag does, and drive the coil/pitch modulation depth below 0.10 to locate where ripple stops improving. None of those goals fully landed, and I want that on record.

The winner (#1, score 0.971) is a 5×2400-turn AWG16 array, 80 mm bore, 90 mm spiral pitch, slurry at φ=25%, remanence 65% giving M≈61.8 kA/m and B≈596 G. It puts 176.1 V into 66 Ω at 11.1 Hz for 176.44 W net. Read the budget honestly: electrical gross is 287.9 W, and I²R in the copper eats 111.5 W — 39% of everything the coil generates is dumped as heat in 3167 m of wire at 41.71 Ω. That is the dominant loss by four orders of magnitude; tube-wall eddy is 2.8e-7 W, utterly negligible, and the hydraulic input to shove the slurry through is only 3.56 W. Mechanical input totals 291.5 W. So the "60.5% efficiency" is gross-generated-into-load divided by mechanical drive — it is NOT overunity and must never be quoted as such. The Lenz reaction force is the real cost of extracting current and it is buried inside that 291 W mechanical figure; the pump does not get the medium back for free.

Modulation depth 0.993 is why it works at all: a uniform magnetized column sweeping past the pickup produces dΦ/dt≈0 and generates nothing. The 0.07 coil/pitch spatial modulation is what creates the flux ripple the coil actually rectifies. Push that toward zero and output collapses — that is the physics, not a knob to minimize blindly.

What's still wrong: velocity never left laminar (Re≈1824), so the EMF-vs-hydraulic scaling test didn't execute — flow stayed pinned at 1.00 m/s across the whole top five. And I²R remains the wall; AWG16 wasn't heavy enough.

Next session: fix gauge at AWG14/13 and re-sweep turns downward to trade voltage for resistance; independently sweep velocity 1.0→2.5 m/s to actually cross Re into transitional flow and measure whether EMF outruns ΔP; and bracket modulation depth 0.02–0.12 to map the diminishing-ripple onset that this run never reached.