Seastead — Tide-Compensating Tension-Leg Winch Design Note

1 · Requirements (from your design)

Parameter	Value	Driver
Working tension	2,000–3,000 lbf per corner	~1 ft pull-down target; pair of helical screws per corner shares the uplift
Travel (cable paid in/out)	≥ 3 ft now; ideally 10–15 ft for future big-tide sites	Tidal range + pull-down + trim + re-termination
Line speed	Tide-following: ≤ 0.5 in/min (even a 10 ft tide). Setup/recovery: 2–5 ft/min	Semi-diurnal tide ≈ 12.42 h sinusoid; max rate ≈ πR/T
Noise	Near-silent — people sleep on the platform; winch is structure-mounted	Liveaboard comfort
Control	Computerized, closed-loop on tension; never allow slack	Tension legs must stay taut through tide + waves
Power / failure	Must hold tension with zero power; per-leg independent supply	Matches your 3-inverter / 3-battery-bank architecture

2 · Short answer

Yes — a slow-speed, self-locking worm-gear drum winch (a small marine "constant-tension mooring winch," the same family used for ship-to-shore gangways), driven by a small brushless DC gearmotor through a worm + planetary reduction, spooling single-layer Dyneema on a large drum, with an in-line load pin for closed-loop tension control.

The single most important feature is the self-locking worm gear: it holds 3,000 lbf with zero power, zero brake, and zero drift, and it is naturally silent at a fraction of an RPM. Pair it with a short elastomer pendant (Seaflex-type) between winch and anchors so the winch only has to follow the slow tide — waves are absorbed passively, and tension survives a power outage.

Why the worm-drive drum winch fits each requirement

Self-locking (non-backdrivable): the load cannot spin the drum. No brake to clunk, no pawl to ratchet, no power to hold. Fail-safe in a blackout.
Enormous reduction in one quiet stage: 40:1–60:1 from the worm alone, plus a planetary stage for 300:1–900:1 total. A small, fast, smooth BLDC motor does the work — small fast motors are far quieter than big slow ones cogging at low RPM.
Creeping is native: at tide-following speed the drum turns ~0.01 RPM. Nothing about that makes noise.
Drum = constant radius: with a single layer of rope, line speed and motor-current-to-tension stay linear, which makes the control loop simple and stable.
Synthetic line (Dyneema), not wire: no wire-rope-on-drum groan, no corrosion, no snap-recoil hazard, gentle on an anodized drum.

3 · Arrangement

One winch per corner (3 total). Winch sits on the deck edge / walkway bracket — not on the living-space wall — on elastomer isolation mounts, powered from that leg's own inverter, matching your triple-redundant power architecture.

4 · Sizing math (estimates — verify before ordering)

Tide-follow rate, 3 ft tide

≈ 0.15 in/min

πR/T, T = 12.42 h sinusoid

Tide-follow rate, 10 ft tide

≈ 0.5 in/min

still trivially slow

Drum speed while tide-following

~0.01 RPM

12 in drum — effectively stationary

Mechanical power, tide-following

< 10 W

3,000 lbf × 0.5 in/min

Drum torque @ 4,500 lbf design pull

≈ 2,250 ft·lbf

on a 12 in (1 ft radius) drum

Total reduction

300:1 – 900:1

worm (self-locking) + planetary or cycloidal

Motor

24/48 V BLDC, 0.75–1.5 kW

intermittent duty; ~700 W if 2.5 ft/min recovery is OK

Line

3/8 in Dyneema

MBS ≈ 19,000 lbf → SF ≈ 6 at 3,000 lbf; D/d = 32:1 on 12 in drum

Line storage

30–60 ft, single layer

10 wraps ≈ 31 ft on 12 in drum; use 14–16 in drum for mega-tides

⚠ One consistency check on your pull-down numbers

You stated ~27,500 lbf displacement and "1 ft ≈ 1/7 of buoyancy" → waterplane ≈ 3,900 lbf/ft total, i.e. ~1,300 lbf per corner per foot. So:

If 2,000–3,000 lbf is per corner (3 × 2,500 = 7,500 lbf), the platform pulls down ~1.9 ft, not 1 ft — check freeboard and ladder/boarding heights.
If you truly want 1 ft of pull-down, set ~1,300 lbf per corner (~3,900 lbf total).

Either works — just pick the number deliberately. Also confirm your helical screws' uplift rating (helical anchors are usually rated for tension, but verify soil assumptions per site).

5 · Control strategy — lazy by design

The trick to silence is to rarely move. Tide is 99% predictable, so feed it forward from tide tables plus a submersed pressure sensor, and let the load pin only trim:

Deadband + dwell: only creep when tension drifts outside ±400 lbf of target for 20+ minutes. Waves and wakes never trigger motion — the elastomer pendant handles those.
Soft ramps: current-limited acceleration over ~60 s. No clunks, no jerk, no structure-borne thumps.
Level keeping: the computer trims per-corner targets ±10% using your platform IMU/depth sensors so the deck stays flat as tension redistributes.
Run in short bursts: with 300:1+ reduction, the motor is smoother doing a 60-second burst at a few hundred RPM every quarter hour than crawling continuously at 5 RPM.

# Per-corner tension controller (runs on that leg's controller / inverter)
TARGET   = 1300 lbf      # per corner for ~1 ft pull-down (see sizing note)
DEADBAND = 400 lbf       # hysteresis band — waves must NOT trigger motion
DWELL    = 20 min        # drift must persist before the winch moves
RATE_MAX = 6 in/min      # hard software limit (tide uses far less)
RAMP     = 60 s          # soft start / soft stop

every 1 s:
    T     = load_pin()
    sched = tide_feedforward()                 # tide table + pressure sensor

    if T < TARGET - DEADBAND and persists(DWELL): creep(+sched)   # tide falling → winch in
    if T > TARGET + DEADBAND and persists(DWELL): creep(-sched)   # tide rising  → pay out

    if T < 500:  alarm(); creep(+2 in/min)     # slack event — correct now, log it
    if T > 6000: alarm(); stop(); drum_lock()  # overload — mechanical fuse territory

6 · Quiet engineering checklist

At the winch

BLDC motor (no brush noise), PWM carrier ≥ 18–20 kHz — above hearing.
Worm in an oil bath + helical planetary (never spur gears).
Dyneema on a smooth drum — no wire-rope groan.
No pawls, no ratchets, no motor brake — the worm is the brake.

Structure-borne (the one that matters)

Elastomer isolation mounts (Sorbothane/rubber) under the winch frame.
Mount on the deck edge / walkway bracket or leg top, never on the 7 ft living wall, which acts as a sounding board.
Flexible couplings where conduit meets the winch.
Deadband control → in a normal month you'll hear it a few seconds a day, if at all.

7 · The elastomer pendant — your "never go slack" insurance

Put a 20–40 ft elastomer pendant (Seaflex-type units, or rubber mooring donuts in series) between the winch line and the delta plate:

Waves and wake: absorbed passively in stretch — zero winch activity, zero noise, zero wear.
Blackout ride-through: with ~10% working stretch, a 30 ft pendant carries ~3 ft of static extension. Tide rises 1–2 ft with no power → tension decays but stays positive for many hours while the worm holds position.
Slack-snap protection: if a big heave momentarily unloads a leg, the pendant keeps the line taut instead of letting it go dead and snap-load — the classic tension-leg failure mode. Your heave plates damp the motion; the pendant removes the consequence.

This is why the control loop can be lazy: the winch follows the tide, the pendant follows the waves.

8 · Alternatives considered

Type	Quiet	Holds w/o power	Verdict
Self-locking worm-drive drum winch + load pin	Excellent	Yes (inherently)	★ Recommended — silent, fail-safe, cheap, exactly matched speeds
Harmonic or cycloidal drive winch	Excellent	Cycloidal mostly; harmonic needs brake	Great precision, pricier — good upgrade path
COTS constant-tension gangway winch (electric)	Good	Usually brake-held	Purpose-built for exactly this job; verify noise & integration
Pneumatic winch (air motor)	Fair (exhaust hiss; muffle it)	No	Elegant constant tension (stall = pressure setting), but compressor noise + new system — skip
Hydraulic winch	Poor (pump)	No	Overkill; pump noise and maintenance on a liveaboard
Planetary drum winch (no worm)	OK	Backdrivable → brake needed	Brake engage/release clunks — reject
Linear (ball-screw) actuator	Good	Some	3+ ft travel = huge actuator; only suits a short-trim 2:1 purchase
Capstan / traction winch	Good	No	For hundreds of feet of cable — overkill here
Counterweight tensioner	Silent	Yes	Perfect constant tension, but 3 × 3,000 lb of deadweight destroys your weight budget
Turnbuckle (manual)	Silent	Yes	Keep one per corner as the storm/emergency back-up — free and silent
Elastic mooring alone	Silent	Yes	Handles ≤ ~3 ft tides alone; no automation for big tides — use as the pendant, not the system

Avoid: anchor windlasses, wakeboard/boat drums, and trailer winches — spur gears, loud freewheels/sprag clutches, and brakes designed for shock loads, not creep.

9 · Safety & redundancy

Shear pin in the drive at ~6,000 lbf — mechanical fuse before line or structure is at risk.
Spring-pin drum lock (servo-engaged, silent) for storm mode and maintenance.
Manual crank on the extended worm input shaft — tedious by design (300:1), but it can always slack or tension a line with no power.
Per-leg independence: each winch on its own leg's inverter/bank — one dead leg never disables the other two mooring controls, mirroring your thruster power architecture.
Software caps: never command more than ~1 ft of pull-down (~1,300 lbf/corner) without explicit operator confirmation; hard stop at 6,000 lbf.
Alarms: tension < 500 lbf (slack risk), tension > 6,000 lbf (overload), motor over-current, load-pin fault → each individually logged.
Environment: IP66/67 motor, 316 stainless shafting, anodized drum, sealed bearings with grease points, marine-rated IP68 load pin (0–10,000 lbf).

10 · Test roadmap (matches your plan)

Anguilla (micro-tide, < 1 ft): run in near-pure feed-forward mode; the winch barely moves. Validate load pins, pendant stretch, noise (sleep aboard with a dB logger on the deck above each winch).
3 ft tide site: enable closed-loop trim; verify slack never occurs across spring-neap cycles, including simulated inverter failures.
Big-tide sites (8–15 ft): only two things change — drum capacity (14–16 in drum, still single-layer) and pendant length. Same winch, same force, same control code.