Active Vertical-Thruster Stabilization
for the Tri-Leg Seastead

Engineering assessment: 3 × 2.5 m submersible mixers (7,000 N each) used for wave-motion control at anchor, plus a convertible mount for dual stabilization / propulsion duty.

First-pass estimatesCaribbean operating areaConcept screening

1. Summary — direct answers

The short version

Yes, the concept works — but as a damper and leveler, not a wave canceller. With well-tuned control, expect roughly 40–60% reduction of the motions people actually feel (heave jerk, chop response, deck tilt) in sheltered-to-moderate Caribbean conditions, up to ~70% for tilt alone. In open roadsteads the reduction drops to ~15–35% because the actuators saturate. Against long ground swell the reduction is near zero — and doesn't need to be anything more, because long swell is gentle.
Why the limit: your three pods total 21 kN ≈ 17% of displacement. That is enough to "hold still" against only about ±0.43 m (±1.4 ft) of water-level variation, and to impose about 0.1 g of corrective acceleration. Seas bigger than that win the quasi-static fight. Spending force on damping (proportional to velocity) buys far more comfort per kilowatt than spending it on cancellation.
Residual motion in a typical 3-ft trade-wind lee: heave about ±0.2 m (±8 in), deck tilt < 0.5°, vertical acceleration ≈ 0.1 m/s² rms. That is ISO-2631 "not uncomfortable" — comparable to a large, stable catamaran tied to a dock on a breezy day. Cooking, sleeping, and walking are normal; seasickness risk is low.
The convertible mount (curved track + winch) is practical in principle — it belongs to the same family as retractable azimuthing thrusters and swing-up saildrives. The hard parts are underwater sealing, biofouling of the track, and holding position with zero backlash under an oscillating 7 kN load. Use hard mechanical locks at discrete angles (90° / 45° / 0°) and never let the winch carry working loads.
A fixed 45° mount is a legitimate v1. It keeps 71% of both capabilities, eliminates the mechanism risk entirely, adds underway ride control and true astern thrust (which your fixed forward-only rim drives lack). Upgrade to the track later if the mission demands it.
Power: damping-mode average draw of 3–6 kW fits your solar + battery budget when used selectively. Continuous full-authority operation (15–25 kW) does not — schedule it, don't live on it.

Everything below shows the arithmetic behind these statements. All numbers are first-pass estimates (±30% is honest) intended for concept screening — validate with a model test before committing steel.

2. Platform numbers this assessment assumes

Parameter	Value	Basis / note
Displacement Δ	12,470 kg (27,500 lb)	Your rated buoyancy at waterline
Weight	122 kN	—
Waterplane area A_wp	≈ 4.8 m² (52 ft²)	3 foil sections at the waterline (NACA 0035 area ≈ 0.685 × t × c, t ≈ 2.98 ft, c = 8.5 ft → ≈ 17.3 ft² each). Consistent with your "1 ft of water level ≈ 1/7 of buoyancy" (computes to 1/8.3 — same ballpark).
Heave stiffness k_z	≈ 48.5 kN/m (3,330 lb/ft)	ρg·A_wp
Effective heave mass m+a_z	≈ 22–30 t	Hull mass + added mass of legs and heave plates (plates assumed ~8×8 ft class). Drives the heave natural period.
Heave natural period T_n	≈ 4.2–4.9 s (nominal 4.5 s)	T_n = 2π√((m+a)/k). Note: bigger/deeper heave plates push T_n up toward the trade-wind sea band — a tuning knob.
Heave T_n with tension legs	≈ 3.6 s	Tethers add ≈ 27 kN/m if pretension ≈ 20% of displacement over a 3-ft pull-down
Roll/pitch natural period	≈ 3.0–3.5 s (≈ 2.1 s tethered)	I ≈ 3–4.5×10⁵ kg·m²; k_θ = ρg·I_wp ≈ 1.5 MN·m/rad
Vertical actuator authority	3 × 7.0 kN = 21 kN (4,700 lbf)	= 17% of vehicle weight
"Hold-still" heave limit	ΣF / k_z ≈ ±0.43 m (±1.4 ft)	Quasi-static water-level variation the pods can resist
Acceleration authority	≈ 0.7–1.7 m/s² (0.07–0.17 g)	21 kN ÷ (12.5 to 30 t), depending on added mass
Differential moment authority	≈ ±108 kN·m ≈ ±4° static	Front pod up + two aft pods down (7 kN × 7.74 m + 14 kN × 3.87 m)
Pod power (estimate)	P ≈ 1.3 kW × (T / 2 kN)^1.5	Actuator-disk model with ~65% efficiency → ≈ 8.5 kW per pod at full 7 kN

k_z = ρ g A_wp ≈ 48.5 kN/m • T_n = 2π √( (m + a_z) / k_z ) ≈ 4.5 s • Δz_hold = ΣF_max / k_z ≈ 0.43 m • P_pod ≈ T^3/2 / ( η √(2ρA_disk) )

3. The physics: what 21 kN of vertical thrust can and cannot do

Fig. 1 — Stabilizer pods at the three corners, each able to push up or pull down with 7 kN. Differential use (one up, two down, etc.) gives roll/pitch leveling authority of about ±108 kN·m.

3.1 Three ways to spend force — only one of them is cheap

Strategy	Force needed (example)	Your authority	Verdict
Hold still (cancel water-level variation, quasi-static)	ρg·A_wp·ζ = ±24 kN in a 3-ft @ 6.5 s sea; ±44 kN in a 6-ft @ 8 s swell	21 kN	Marginal in 3-ft seas (~86% of amplitude, no margin for dynamics); impossible in swell
Fight acceleration (impose counter-acceleration)	m_eff × a ≈ 7–15 kN for the accelerations in 3–6 ft seas	21 kN	Yes in moderate seas, partial in big ones
Damp velocity (active "sky-hook" damper, force ∝ velocity)	6–12 kN to roughly halve resonant heave (velocities are high at resonance, so force is efficient)	21 kN	Yes — best comfort per kilowatt
Level the deck (differential roll/pitch moments)	≈ 25 kN·m per degree of correction	±108 kN·m	Easy — hold < 0.5° almost always

Core design principle Damping is cheap; cancellation is expensive. A velocity-proportional force only does work on motion you actually remove, and near resonance the platform moves fast, so modest forces buy large reductions. Holding the deck at fixed elevation while the sea swings ±0.5 m means fighting ±24 kN continuously — most of your total authority, burned as heat in the water. So spend authority in this order:

Kill resonant ringing (heave near ~4.5 s, roll near ~3 s) — highest payoff.
Keep the deck level (< 0.5°) — cheap, huge livability gain.
Trim slow offsets (mean heave, list from consumables, tether tension assist).
Only then attempt partial wave cancellation in the 5–8 s band, and let long swell pass underneath.

3.2 The swell you can't cancel — and why that's fine

A 6-ft, 13-second swell lifts the water ±0.9 m over many seconds. Holding your deck motionless through it would require ≈ ±44 kN of slowly-varying force — twice your total authority — indefinitely. Don't try. The good news: a 13 s motion produces tiny accelerations (≈ 0.1–0.2 m/s² even at full amplitude), so the platform simply "breathes" a few feet, slowly and gently. Nobody gets seasick at 0.08 Hz with 0.07 m/s² rms; you just design for it (long mooring lines, dinghy painter slack, boarding ladder that tolerates ±1 m). The same logic applies to the tension legs: 3 ft of pull-down adds stiffness, but flattening a 1 m swell quasi-statically would need ≈ 48 kN of extra tether force — that would overload the helical screws. Let the slow motion happen; actively manage only the fast, uncomfortable part.

4. Expected performance by sea state

Five representative Caribbean conditions, from a glassy lee to open-roadstead chop. "Baseline" is the passive platform (heave plates fitted, no active control); "Active" assumes well-tuned control with saturation-aware force allocation. All values are amplitude (±) or rms, as noted.

ID	Condition	Sea state
A	Sheltered lee, light chop	Hs 0.5 m (1.6 ft) @ 5 s
B	Typical trade-wind lee	Hs 1.0 m (3.3 ft) @ 6.5 s
C	Open roadstead, fresh trades	Hs 1.8 m (6 ft) @ 8 s swell + 0.6 m @ 5 s chop
D	Long ground swell	Hs 2.4 m (8 ft) @ 13 s
E	Steep wind-against-current chop (near heave resonance)	Hs 1.2 m (4 ft) @ 4.5 s

4.1 Baseline (passive) motion

ID	Heave amplitude ±	Vertical accel (rms)	Deck tilt
A	0.16 m	0.06 m/s²	0.4°
B	0.34 m	0.22 m/s²	0.8°
C	0.85 m (swell-dominated)	0.55 m/s²	1.3°
D	1.05 m (follows swell)	0.07 m/s²	0.6°
E	0.75 m (resonant amplification ~1.25×)	1.0 m/s²	1.0°

4.2 With active stabilization

ID	Reduction: heave / tilt / accel	Residual heave ±	Residual accel rms	Residual tilt	Avg. electrical power
A	55% / 65% / 55%	0.07 m	0.03 m/s²	0.15°	1–2 kW
B	40% / 60% / 45%	0.20 m	0.12 m/s²	0.3°	3–5 kW
C	15% / 60% / 35%	0.72 m	0.36 m/s²	0.5°	8–14 kW (scheduled)
D	~5% / 50% / ~5%	1.0 m	0.07 m/s²	0.3°	0–2 kW
E	45% / 55% / 45%	0.45 m	0.55 m/s²	0.45°	6–10 kW

Reading the table: in B and E the pods are doing exactly what they're good at — eating resonant and short-wave motion. In C the 8 s swell component is quasi-static and mostly passes through untouched (that's the difference between 15% heave reduction and 60% tilt reduction); the pods still remove most of the chop's "bite." In D there is nothing worth fighting.

Fig. 2 — Where the pods earn their keep: tilt everywhere, heave/acceleration in conditions A, B, E; little to gain in C's long swell or D.

Fig. 3 — Illustrative heave transfer functions. Active control flattens the resonant hump and digs into the 5–8 s trade-wind band; curves converge at long periods where no one has the force to (or needs to) fight the swell.

4.3 Graceful degradation (what happens when you run out of force)

In bigger seas the pods will saturate. Design the allocator with a fixed priority so behavior degrades predictably instead of oscillating:

Priority 1 — heave-rate damping (all three pods, symmetrical): never sacrificed.
Priority 2 — attitude leveling (differential): sacrificed first if starved.
Priority 3 — mean-heave trim / tether-tension assist (slow integrators): tiny force, always affordable.
Priority 4 — horizontal station-keeping thrust (convertible mount only): takes whatever is left inside the per-pod thrust circle T_v² + T_h² ≤ T_max².

The practical consequence: in a fresh blow you automatically revert from "wave cancellation" to "damper + leveler," which is precisely the mode that is cheap to run and sufficient for comfort.

5. What it feels like: comfort assessment

Vertical accel (rms, ISO 2631-1 weighted)	Comfort label	Who lands here
< 0.315 m/s²	Not uncomfortable	A, B, D (active and passive); C and E with active control at the better end
0.315 – 0.63 m/s²	A little uncomfortable	C active (0.36), E active (0.55); C passive (0.55)
0.63 – 1.0 m/s²	Fairly uncomfortable	E passive (~1.0) — the case active control exists for
1.0 – 1.6 m/s²	Uncomfortable	Worse chop than E; outside design envelope

Motion sickness

Susceptibility peaks for vertical oscillation around 0.15–0.25 Hz — unfortunately exactly your heave band. However, published motion-sickness incidence data suggest symptoms become common only above roughly 0.3–0.5 m/s² rms in that band. Your controlled residuals (0.03–0.36 m/s² rms in A–C, E) sit at or below that threshold: expect a small fraction of unadapted guests to feel queasy on a fresh day in condition C, and essentially nobody in A/B. Long swell (D, 0.08 Hz) falls below the susceptible band — it causes visual mismatch complaints at most, mitigated by keeping a horizon view through the windows.

Plain-language descriptions

A — sheltered lee: Dock-like. Coffee survives. You notice motion only by looking at the waterline.
B — typical trade-wind lee: A gentle, slow sway — like a large stable catamaran tied to a dock on a breezy day. Walking, cooking, sleeping, and electronics all normal. This is your bread-and-butter "great stability in the lee of an island" condition, and the tether + thruster combination is at its best here.
C — open roadstead: You'll know the ocean is there: a slow ±0.7 m lift-and-settle from swell that no affordable machine will remove, but with the chop's sharp accelerations largely deleted. Comparable to a stable dive boat at anchor. Meals OK with care; secure loose objects.
D — ground swell: A slow, majestic breathing of several feet over ~13 s. Effortless physically; the design tasks are operational (slack in lines, dinghy painter, ladder geometry), not dynamical.
E — steep chop: The scenario that would make the passive platform genuinely unpleasant (near-resonant heave, ~1.0 m/s² rms) becomes merely lively (~0.55). This is where active stabilization most obviously "earns its cost."

Secondary dividends: lower accelerations mean lower cyclic loads on the walkway bolts, railing, dinghy ropes, and furniture — and the two-seastead connecting walkway becomes dramatically more usable, since both controllers can co-operate to minimize relative motion at the mating points.

6. Recommended control architecture

Loop	Feedback	Bandwidth	Notes
Heave-rate damping (priority 1)	IMU vertical velocity (filtered accelerometer + RTK-GNSS)	0.3–1 Hz	"Sky-hook" damper; the workhorse. Lag-tolerant — degrades gracefully if pods are slow.
Attitude leveling (priority 2)	RTK-GNSS / IMU roll & pitch	0.1–0.3 Hz	Differential pod commands; hold < 0.5°; authority ±4° static.
Mean trim (priority 3)	Slow integrators on heave offset and list	~0.01 Hz	Compensates fuel/water/battery-state changes; prevents wind-up.
Tether-tension assist	Load cells on the three mooring units	~0.1 Hz	Push down as a crest approaches to keep tethers from going slack — protects the helical screws from shock loads.
Wave feedforward (optional but valuable)	Small upward-looking sonar / wave gauges on the leading edges of the legs	predictive	You already plan to head into the seas — the three leading edges are perfect sensor perches. Feedforward buys back much of the phase lost to pod lag.

Sensors: RTK-GNSS (heave/tilt truth), 100+ Hz IMU per corner, pod thrust/temperature monitoring, optional wave profiler. Everything triple-redundant to match your per-leg power philosophy — each pod fed from its own leg's inverter/battery, so one leg's electrical failure costs one pod, not three.
Latency budget: see §7. Total loop delay must stay under ~1 s; use Model-Predictive Control or a Smith-predictor structure to explicitly compensate pod dynamics.
Fail-safe: on any fault, pods command to zero thrust and the platform reverts to purely passive behavior. This is acceptable because the passive platform is already comfortably stable in A/B/D — active control is an enhancement layer, never a stability-of-life dependency.

7. Hardware realities, risks, and the latency problem

7.1 Thrust-reversal latency — the make-or-break spec

A 2.5 m mixer is a big, slow-spinning machine (order 60–90 rpm at full thrust). Rough numbers: rotor inertia J ≈ 60–120 kg·m², drive torque ≈ 1,100–1,200 N·m at rating. A full thrust reversal means slewing the rotor through zero:

t_rev ≈ J·Δω / τ_drive + slipstream reversal lag ≈ 0.7–1.4 s + 0.2–0.5 s ≈ 1–2 s

Against a 6.5 s wave (quarter-period 1.6 s) that is tight but workable — if you engineer for it:

Specify regenerative VFDs with ~2× overload torque; recover braking energy into the batteries.
Keep the props always spinning at a small bias so you modulate around a live plant instead of starting from rest (idle spin costs < 1% of rated power).
Bench-test one pod's step-thrust response before committing — this single measurement decides how aggressive the control can be. If reversal proves slower than ~1.5 s, remedies are: controllable-pitch props (instant thrust reversal, expensive), twin counter-rotating props per pod, or accepting a damping-only control law (still delivers most of the comfort benefit).
Smaller/faster pods (e.g., 1.6–1.8 m) would trade peak thrust for response speed — worth a sensitivity study, since responsiveness buys more comfort than peak force in the 4–7 s band.

7.2 Other engineering realities

Fatigue: ±7 kN at 0.1–0.3 Hz is ~3–8 million cycles/year. Detail the pod brackets for 10⁸ cycles (machined fittings + bolts preferred over welds at the hot spots).
Fouling: warm Caribbean water grows on everything in weeks. Props tolerate it; tracks and sliding surfaces do not (see §8). Budget diver cleaning or growth-release coatings.
Safety: a 2.5 m propeller spinning at chest-depth below the platform is a genuine hazard for swimmers and divers. Guards cost thrust; instead use strict lockout procedures, obvious markings, and automatic shutdown on "person in water" flags. Also: the downwash jet (~1.7 m/s induced velocity) will stir sediment in water shallower than ~6 m — in skinny anchorages, lean on the tethers and keep the pods quiet.
Noise/vibration: disk loading is very low (~1.4 kPa) and tip speeds ~12 m/s, so radiated noise is modest — neighbor- and reef-friendly. Mount pods on elastomers to keep structure-borne hum out of the living space.
Pod ↔ heave-plate interaction: keep ≥ 1 prop-radius clearance below the lowest heave plate so the discharge jet doesn't recycle through the plate (it would corrupt both your damping and your plant model).
Electrical: 3 × ~9 kW peak of new drive load. Your per-leg inverter/battery independence maps perfectly onto one pod per leg — preserve that mapping in the wiring design.

Loop element	Typical delay
Sensing (IMU/GNSS) + filtering/estimation	30–80 ms
Control computation	5–10 ms
VFD torque response	30–80 ms
Rotor speed change (J·Δω/τ)	300–900 ms
Slipstream/thrust reversal	200–500 ms
Total	≈ 0.6–1.5 s (vs 1.4–1.8 s quarter-period at 6–8 s waves)

8. The convertible mount: track, winch, and the 45° question

Fig. 4 — One leg's convertible pod: vertical for stabilization, horizontal for propulsion / station-keeping, 45° for split duty. Same picture applies at all three corners.

8.1 Is it practical? Yes — with eyes open

This mechanism belongs to a well-precedented family: retractable azimuthing thrusters, swing-up saildrive legs, drop-in tunnel thrusters, and yacht daggerboard/trunk systems all move heavy hardware under load in seawater. Nothing here violates known practice. The difficulty is moderate-to-high, concentrated in four places:

Zero-backlash positioning under an oscillating 7 kN load. A winch-held position will hunt and chatter as thrust reverses at wave frequency. Remedy: hard mechanical locks (spring-loaded pins) at 90°, 45°, and 0°. The winch only transits between detents; it never carries working load. If you truly need continuous in-between angles, use a self-locking screw jack with a brake, not a rope winch.
Underwater sealing and fouling. Pivot shafts need lip seals + labyrinths; tracks foul fastest of all. Consider enclosing the rail in a slotted tube, specifying growth-release coating, and designing the pivot so the pod can be unbolted and hoisted to the walkway for service without divers.
Structural load path. The track/bracket must react 7 kN at any angle plus dynamic overshoot (design 2×), into a leg shell that is also carrying heave plates and battery mass. Add internal ring frames at the attachment; this is also a fatigue hotspot (§7).
Hydrodynamics of the horizontal mode. Mounted aft of the trailing edge at ~5 ft depth, the pod enjoys straightened flow and good immersion — good. Expect some thrust-deduction from the leg's wake and slight asymmetric inflow in turns; minor. In the 45° fixed case, add a shaft brake so the props don't windmill (drag) during transits, and fair the brackets.

8.2 Option comparison

Criterion	A — Fixed 45° brackets	B — 3-position track + locks	C — Dedicated vertical pods
Stabilization authority	●●●○○ (5.0 kN vert./pod)	●●●●● (7.0 kN)	●●●●● (7.0 kN)
Propulsion / astern / station-keeping	●●●○○ (always available)	●●●●● (full 7 kN axial)	●●○○○ (none; rim drives only)
Simplicity & reliability	●●●●●	●●○○○	●●●●○
Maintenance ease	●●●●○	●●○○○	●●●●○
Cost efficiency	●●●●●	●●○○○	●●●●○
Underway ride-control bonus	Yes (built in)	Yes (at 45° detent)	No
Verdict	Recommended v1	Phase-2 upgrade if missions demand full authority in both roles	If you'd rather not touch propulsion at all

8.3 The 45° compromise, quantified

Per pod: 4.95 kN vertical + 4.95 kN horizontal (71% of each axis's dedicated capability).
Fleet totals: 14.9 kN vertical → hold-still limit shrinks from ±0.43 m to ±0.31 m — still covers conditions A and B comfortably. 14.9 kN horizontal → holds station against roughly a 30-knot wind on your ~75 m² of windage (≈ 11 kN at 30 kt), with the tethers doing the steady-state work anyway.
Bonus you didn't ask for but will like: at 45° the pods work while underway — the vertical component becomes an active ride-control system for passages between anchorages, and the horizontal component supplements the six rim drives. And because your rim drives are fixed-forward-only, the pods give you your only real astern thrust (~15 kN) — a meaningful handling upgrade for docking and dinghy operations.
Controller note: with fixed 45° units, the two axes share one saturation ellipse (T_v² + T_h² ≤ T_max²). The allocator must respect it — a standard constrained-allocation problem, solved identically to the priority scheme in §4.3.

Recommendation Build v1 with fixed 45° mounts: zero mechanism risk, always-ready in every mode, and it validates the entire control system, the power budget, and the comfort gains. Fit one prototype convertible (track-and-lock) pod on one corner as a technology demonstrator; if it survives a season of fouling and 10,000 duty cycles, convert the other two. You lose almost nothing now and keep the upgrade path open.

9. Energy budget — "use it more or less depending on your solar"

Exactly right instinct. Ballpark numbers for your configuration:

Resource	Estimate	Note
Solar array	~13–14 kWp on ~65 m² usable roof	≈ 60–70 kWh/day in Caribbean sun
Battery bank	~3,100 kg LFP ≈ 340 kWh installed, ~310 kWh usable	25% of displacement, per your layout
Pod power model	P ≈ 1.3 kW × (T/2 kN)^1.5 per pod	8.5 kW each at full 7 kN; 3.2 kW at 3.5 kN; 1.4 kW at 2 kN

Operating mode	When	Avg. draw	Energy per 12 h	Sustainable?
Stabilizers off	Calm / snugly tethered	0	0	Yes
Light damping	Condition A/B days, daytime-biased	2–4 kW	24–48 kWh	Yes
Standard damping	Breezy lee, all day	4–6 kW	48–72 kWh	Yes, with sun
Heavy (roadstead/chop events)	Limited hours, storm-managed	8–14 kW	50–85 kWh per 6 h	Scheduled; recharge after
Continuous full-authority hold	Not recommended	15–25 kW	180–300 kWh	No — exceeds solar; drains bank in ~1 day

Practical policy: let the energy manager gate stabilization on battery state and solar surplus (e.g., full authority above 80% SOC and midday; damping-only otherwise; automatic standby below 40%). Regenerative braking during thrust reversals returns a few percent — take it, but don't count on it. Multi-day overcast spells simply mean reverting to passive mode, which — by design intent — remains comfortable in exactly the conditions where you'd be short on sun anyway (sheltered water).

10. Suggested next steps

Bench-test one mixer pod: measure thrust step/reversal response, P(T) curve, thermal limits. This single dataset calibrates everything above.
Free-running model (1:6–1:8) with scaled vertical actuators: measure baseline RAOs, verify the added-mass/T_n assumptions, tune damping control in head and bow-quarter seas.
CFD / tow tests for pod–heave-plate interaction and bracket drag (especially the 45° fixed option).
Mechanism prototype: one track-and-lock pod on a dock-side test frame; 10,000 articulation cycles + a 3-month fouling trial before trusting it offshore.
Controller HIL simulation against recorded Caribbean buoy spectra (CDIP/NDBC) before first sea trials.
Define the acceptance test now: "≥ 40% reduction in heave rms and ≥ 50% in tilt rms in a 3-ft @ 6–7 s sea, measured by RTK-GNSS, pods averaging ≤ 5 kW." Ship what meets it.

11. Assumptions, sensitivities, and a packing cross-check

All figures are first-pass estimates (±30%): added mass, heave-plate size, pod efficiency, and wave climate are assumed, not measured. The biggest sensitivities: effective heave mass (sets T_n and therefore everything), pod reversal latency (sets how much cancellation survives), and actual solar yield (sets duty cycle).
If heave plates end up larger or deeper than assumed, T_n slides toward the trade-wind band — re-check before freezing plate geometry; the active system can compensate but shouldn't have to.
Results assume competent control implementation (saturation-aware allocation, latency compensation). A naive position-feedback controller will perform worse and can even add motion near resonance.
Heading into the seas helps (feedforward sensors on the leading edges, lower leg loads, dinghy shielding). The triangular layout is reasonably omnidirectional, so wind shifts are tolerable without repositioning.

Packing cross-check (since the container plan drives everything): the nested leg pair stands ≈ 8.5 + ~3 ft ≈ 11.5 ft tall — taller than the 8.9 ft interior, so it only fits if loaded tilted along the 7.7 × 8.9 ft cross-diagonal (≈ 11.8 ft). If that's your intent, it works: pair + third leg lengths 21.5 + 21.5 = 43 ft ≤ 44.6 ft ✓, combined widths ≈ 6.2 ft ≤ 7.7 ft ✓. Worth confirming the tie-down design for the tilted pair before the shipyard visit.