A hydrogen airship you can build in a shop and then never have to land.
Powered by sunlight, fuel cells and engines so that no single technology failure ends the flight. Sized to be lived aboard by two people for a year. Able to land on water and operate as a boat.
This is an open engineering notebook, not a concept renderer. Every number was computed by the model at build time, by the same functions the tests call. Where the model is guessing, it says so. One gate on this site is failing, and it is left failing.
The near half of the hull and cover removed. Cells, keel corridor, gondola and every compartment.
Three findings that decided it
Each came out of building the model rather than out of reading about airships, and each reversed something this project had already written down.
Drawing the arrangement made it 25 m longer
The baseline was 90 m for as long as the mass budget was a fraction. Giving the compartments, the machinery and the array real positions and real masses turned it into a statement, and 90 m came out 4,534 kg heavy. An aeroplane that comes out heavy loses range and still flies. An airship has no such trade: the buoyancy is fixed by the envelope.
A sealed pneumatic float is stiffer than the water
Its stiffness goes with ABSOLUTE pressure and not gauge, so it is nearly sixty times stiffer than the waterplane it replaces. It limits force only if it VENTS. A rigid hull is held to sea state 1; a vented bag reaches 6; a sealed one reaches none at all.
The daily swing is bigger than the design load
Twenty kelvin of superheat moves lift by about two tonnes. The trim the vehicle rests on water at is 800 kg. It floats off its float in the afternoon and presses two tonnes onto it before dawn, every day. No passive device can be sized for that, and this gate is left failing.
The case, in order
Read top to bottom it is the argument for the design. Jumped into anywhere it is a reference.
- 01The shipWhat is it, where is everything, and could you live in it?The arrangement and the interior: a cutaway, an inboard profile, four sections and a plan of every room with what is in it. All drawn from the same stations, footprints and masses the budget integrated.
- 02Why this and not something elseRigid, semi-rigid, non-rigid, hybrid-lift or variable-buoyancy?All five on one basis, each calibrated on a vehicle that flew. Three are lighter than the one chosen, and each is lighter for a reason that costs something a liveaboard cannot pay.
- 03Does the loop close?Can sunlight alone keep it up for a year?Solar collection integrated over the real hull surface, a fuel cell and electrolyzer round trip, and a day-by-day mission integration that reports the day it fails rather than an average that hides it.
- 04Can it be built?Does the square-cube law let a carbon frame carry this?The mass fraction against every rigid airship with published figures, the buckling allowables that actually size the frame, and the gust case that turns out to govern rather than the static one.
- 05Land it on waterCan it float, and can it get anywhere afterwards?Flotation is trivial and it is not the problem. A simulator that integrates the real seakeeping, and the finding that a sealed pneumatic float is stiffer than the water it replaces.
- 06Fly itWhat does it feel like to handle?The project’s own 6-DOF solver at 100 Hz, with the full added-mass tensor. Slow to respond, slow to stop, and overdamped at cruise where it wallows at rest.
- 07Does it hold up?Does the model reproduce ships that actually flew?Every rigid airship with published figures, modelled from its own envelope and compared. Where the model misses, the discrepancy is recorded rather than tuned away.
- 08Where the model is guessingWhat would change the answer if it were measured?The uncertainty register, sorted by how much each unknown moves the endurance number. This is the research queue, and one gate on this site is failing on purpose.
The rules everything follows from
There is one model. The site reads it, the simulator integrates it, the documentation renders from it. If a number appears in prose and also in the solver, that is a bug.
The corollary that governs day to day work is that the figure of merit is days aloft. Not speed, not range, not payload. When two designs conflict the one that stays up longer wins, and it is why hybrid lift loses here and wins nearly everywhere else.
Never invent a number. A lint rule fails any numeric literal in the physics packages that lacks a source or a derivation. Values that are genuinely unknown are encoded as uncertain, with a reason and with what would resolve them, and they appear in the research queue sorted by how much each one moves the endurance figure.
The baseline is stationed at 15° latitude and 2,000 m, holding station against 8 m/s for 65% of the day and drifting the rest.