A general arrangement drawing is the ship's constitution — every subdivision boundary, escape route, tonnage envelope and sightline from the bridge has to satisfy a rule before it earns its place on the plan. This chapter works through the reasoning and the numbers behind each of those constraints.
A general arrangement is not drawn freely from the bow aft — the internal geometry of a ship is very largely fixed before the accommodation and cargo spaces are ever laid out, because subdivision is a damage-stability problem long before it is a layout problem. The naval architect's job is to place watertight bulkheads so that the ship survives the flooding scenarios the rules require it to survive, and only then fit the rest of the plan around those fixed boundaries.
Two related methods drive this: the older deterministic floodable-length approach and the probabilistic damage-stability index used more widely today. Both start from the same physical question — if a given compartment, or combination of compartments, floods completely, does the ship stay upright and afloat with an acceptable margin? — and both convert the answer into a set of permitted bulkhead spacings along the length of the ship. The designer does not choose bulkhead positions and then check them; the check effectively generates the positions.
One bulkhead is never left to the designer at all: the collision bulkhead. Because a collision is statistically the flooding event most likely to happen right at the bow, its position is fixed by rule within a band measured aft of the forward perpendicular, with an allowance made forward if the hull carries a bulbous bow extending beyond the perpendicular. A junior engineer who "optimises" this bulkhead aft for more forecastle stowage, or forward for more accommodation, has not optimised anything — they have produced a non-compliant drawing.
Bulkhead positions are an output of the damage-stability calculation, not an input to it. Draw the subdivision after the stability case is solved, not before.
The floodable length at a point x along the ship is the greatest length of compartment, centred on x, that can flood without immersing the margin line — a reference waterline used specifically for this check, not the ship's actual deepest operating waterline. Plot that maximum permissible length at every point along the hull and the result is the floodable length curve; the actual bulkhead spacing chosen for the ship has to sit on or inside that curve everywhere.
The margin line is deliberately not the bulkhead deck itself; it sits a short distance below it, so the check retains a margin against the deck itself becoming awash. A resulting waterline that just touches the margin line after flooding is a pass at the limit, not a comfortable pass — and the whole exercise assumes the ship starts from an intact, permissible loading condition, not an overloaded one.
Permeability is the part of the calculation students most often skip past. A flooded compartment does not fill completely — cargo, machinery, furniture and structure all displace water — so each space type is assigned a permeability factor representing the fraction of its volume that can actually flood. Accommodation spaces flood almost completely, since they are mostly open air; machinery spaces flood noticeably less, because engines, generators and piping occupy real volume; a loaded cargo hold floods less still. Get the permeability wrong for a space and the floodable length calculated for that section of the ship is wrong in the same proportion — too generous if you overstate it, unnecessarily restrictive if you understate it.
Escape provision is a layout constraint that has to be satisfied at the same time as everything else, not bolted on once the accommodation is drawn. The baseline requirement is two means of escape from essentially every normally occupied space, positioned so that a single fire, flood or structural failure cannot plausibly block both routes at once — which is why the two escapes from a given space are required to be widely separated, not merely two doors off the same alleyway.
The routes themselves have to work as a continuous path, not just as a compliant doorway at each end. Stairways and passageways along the escape route need to stay clear of cargo, stores and fitted equipment, and the route has to lead all the way to an open deck or embarkation position without doubling back through spaces that could themselves be the source of the emergency. A beautifully compliant stairwell that terminates in a space normally kept locked, or that passes through the galley, has not actually delivered an escape route.
A requirement that surprises a lot of candidates is that these routes must remain usable when the ship is heeled, not only when upright at the design waterline. A ladder, stair pitch or door swing that works perfectly on an even keel can become impractical, or physically blocked, once the deck it sits on is tilted — so escape arrangements are checked, at least conceptually, against a heeled condition, not just the calm-water layout drawing.
Tonnage measurement is where a lot of exam candidates instinctively reach for a mass and get the question wrong before they have written a line. Gross and net tonnage are both volumetric — dimensionless numbers derived from enclosed volume, not from displacement or deadweight — and they exist to standardise how a ship's "size" is described for dues, manning scales and regulatory thresholds, independent of what the ship is actually built from or loaded with.
The log term in K₁ is doing real work, not just adding complexity. As enclosed volume grows, K₁ grows too, but only slowly — so gross tonnage grows a little faster than volume alone for very large ships, reflecting that bigger hulls tend to be proportionally more efficiently packed with enclosed space. It is easy to forget that K₁ itself depends on V, and to carry a K₁ computed for one ship's volume into a different volume without recalculating it.
Net tonnage is built from the volume of cargo spaces specifically, with a correction term for the ratio of draught to depth and, on passenger ships, for the berths carried, and it is never allowed to be taken as less than a fixed fraction of the gross figure. It broadly tracks earning capacity rather than overall size, which is why net tonnage — not gross — is what many port and canal dues are actually based on. Both figures trace back to the old "register ton" convention, a volume unit of roughly 2.83 m³, which is why tonnage numbers for a modest coaster can still run into the thousands despite the ship weighing nothing like that in tonnes.
Visibility from the conning position is one of the few human-factors requirements written into the arrangement with an explicit geometric limit, which makes it a favourite for numerical exam questions. The basic requirement is that the sea surface forward of the bow must be visible from the bridge out to a limited distance — obscured no further than two ship lengths, or a fixed absolute distance, whichever is the smaller — so that on a short ship the ship-length rule governs, while on a very long ship the absolute distance takes over.
Checking this during design is mostly triangle geometry. Treat the conning position as an eye at a known height above the waterline, treat the tallest obstruction ahead of it — a forecastle, a stack of deck cargo, a crane — as a point at its own height and horizontal distance from the eye, and extend the sightline from the eye over the top of that obstruction down to where it meets the sea surface. Everything between the ship and that point is invisible from the bridge; move the obstruction further forward, or lower it, and the obscured distance shrinks.
The same visibility case also limits blind sectors caused by masts, kingposts and samson posts to the side and astern of the bridge: any individual sector is kept small, the combined total across all such obstructions is capped at a larger figure, and a clear arc has to be preserved close to dead ahead regardless of everything else on the weather deck. In practice this is what stops deck-cargo stowage plans and mast positions from being finalised purely on a stability or lashing basis — visibility has a vote too, and it is usually the container stack, not the hull form, that has to give way.
Visibility is checked against the ship's actual deepest operating condition and its tallest realistic deck cargo, not the light-ship arrangement drawing — a compliant empty ship can still fail loaded.
The general arrangement drawing itself is the document all of the preceding requirements are reconciled onto: a deck-by-deck plan showing every compartment, watertight boundary, escape route, tank and access, cross-referenced against the capacity plan and the tonnage calculation that depends on it. Spaces open to the weather, and certain excluded volumes, are treated differently in the tonnage sums than fully enclosed spaces — a detail worth checking directly against the arrangement drawing rather than assuming from the deck plan alone.
Habitability standards shape the accommodation block in the same binding way that stability shapes the hull subdivision. Minimum headroom, floor area, ventilation and noise limits for crew accommodation are set out under the maritime labour framework governing seafarers' living conditions, and a compartment that satisfies structural fire protection and escape requirements but falls short on headroom or ventilation is still a non-compliant space. These standards interact with everything else on the drawing: an escape stair widened for compliance eats into an adjoining cabin's floor area, and a cabin relocated to gain deck space can move accommodation into a position that then fails the bridge-visibility check.
None of these requirements are solved in isolation on a real project — subdivision, escape, tonnage, visibility and habitability are worked concurrently, and the arrangement drawing is simply where the compromises between them are made visible. Reading a finished GA drawing well means being able to see which constraint pushed which boundary into the position it occupies, which is exactly the skill the worked examples below are built to exercise.
The three examples below work through a tonnage calculation, a bridge-visibility check and a net-tonnage case with a floor applied — the same reasoning, run with real numbers, that sits behind the qualitative rules above.
A small coaster has four enclosed spaces contributing to her total enclosed volume: a forward cargo hold, an engine-room casing, an accommodation block and the remaining enclosed deck spaces. Determine the ship's gross tonnage, and state whether she falls above or below the 500 GT threshold that changes which convention requirements apply in full.
Forward cargo hold volume = 4200 m³ Engine-room casing volume = 3100 m³ Accommodation block volume = 1800 m³ Remaining enclosed deck spaces = 900 m³
Determine the ship's gross tonnage, and state whether she falls above or below the 500 GT threshold that changes which convention requirements apply in full
Total enclosed volume.
Gross tonnage is built from the sum of every enclosed space on the ship, so add the four volumes before touching the tonnage formula.
Tonnage coefficient.
K₁ depends on this same total volume, so it has to be evaluated for V = 10 000 m³, not carried over from another ship.
Gross tonnage.
Apply the coefficient to the same volume it was derived from.
Compare the result with the 500 GT threshold commonly used as a dividing line for the scope of convention requirements.
AnswerGT ≈ 2800 (dimensionless), comfortably above the 500 GT threshold.
The trap: plugging a K₁ borrowed from a different ship's volume into this GT formula, or reporting the answer in tonnes as though it were a mass.
A container ship, length L = 180 m, has her conning position 21 m above the waterline. A stack of deck cargo 40 m forward of the conning position rises to 15 m above the waterline, and the stem is a further 10 m forward of that stack. Check whether the sea surface obscured ahead of the bow satisfies the visibility limit.
Height of eye above waterline, h_e = 21 m Height of obstruction above waterline, h_o = 15 m Horizontal distance, eye to obstruction, x = 40 m Horizontal distance, obstruction to stem = 10 m Ship length, L = 180 m
Check whether the sea surface obscured ahead of the bow satisfies the visibility limit
Sightline geometry.
The critical line of sight runs from the eye, grazing the top of the cargo stack, down to the sea surface. Similar triangles give the horizontal distance from the eye to where that line meets the water.
Obscured distance ahead of the bow.
The visibility limit is measured from the stem, not from the eye, so subtract the eye-to-stem distance from D.
Governing limit.
The permitted obscured distance is the lesser of two ship lengths and the fixed absolute limit.
Compare the obscured distance with the governing limit.
Answer90 m obscured versus a 360 m limit — the arrangement passes, with 270 m of margin.
The trap: measuring the obscured distance from the conning position instead of from the bow, which understates how much margin the arrangement actually has.
A small general cargo ship has gross tonnage GT = 800, cargo-space volume Vc = 1000 m³, moulded depth D = 20 m and moulded draught d = 13.5 m, and carries no passengers. Determine her net tonnage.
Gross tonnage, GT = 800 Cargo volume, Vc = 1000 m³ Moulded depth, D = 20 m Moulded draught, d = 13.5 m No passenger spaces
Determine her net tonnage
Tonnage coefficient for the cargo volume.
Net tonnage uses its own coefficient, evaluated on the cargo volume Vc, not on the ship's total enclosed volume.
Draught/depth ratio term.
This factor reflects how much of the ship's depth is actually being used to carry cargo at the draught in question.
Raw net tonnage.
With no passenger spaces the berth term drops out, leaving the cargo-volume term alone.
Apply the statutory floor.
Net tonnage is never taken as less than three-tenths of gross tonnage — check the raw figure against that floor before reporting a result.
AnswerNT = 240 (the floor value), not the raw 210.6 the formula alone would give.
The trap: reporting the raw calculated figure of 210.6 without checking it against the 0.30 × GT floor, which is exactly what this ship's numbers are designed to trip.
GT = K₁·V, K₁ = 0.2 + 0.02 log₁₀VV = total enclosed volume, m³; GT is dimensionlessNT ≈ K₂·Vc·(4d/3D)² [+ berth term], ≥ 0.30 GTVc = cargo volume; ratio term capped at 1Collision bulkheadPosition set by rule aft of the forward perpendicular; not a design choiceMargin line: 76 mm below the bulkhead deckThe flooding limit the floodable-length curve is built againstFloodable length L_f(x)Max compartment length at x that floods to the margin line, given permeabilityPermeabilityAccommodation floods most, machinery less, loaded cargo leastTwo means of escapeWidely separated, continuous, and usable when the ship is heeledBridge visibilitySea surface obscured ahead capped at 2L or a fixed absolute distance, whichever is lessBlind sectorsEach kept small, the combined total capped larger, dead-ahead kept clearRegister tonHistorical basis of tonnage: 100 ft³ ≈ 2.83 m³