What the simulator shows
A floating ship is pushed down by its weight at G and up by its buoyancy at B. Upright, the two forces act on one vertical line. Heel the ship and the immersed shape changes, so B moves towards the low side. The horizontal gap between the two forces is the righting lever GZ. While GZ is positive, the ship comes back upright.
The top figure is the midship section drawn to scale, with the waterline, G, B and the initial metacentre M. Drag the hull or use the heel control and the section rotates. The lower chart is the GZ curve for both sides, with the areas of the IMO criteria shaded. "Let it roll" releases the ship from the set angle and lets it swing freely.
Things to try
- GM is not enough on its own. The default ship has GM0 = 0.76 m and passes all six criteria. Raise KG from 7.6 to 8.2 m. GM0 falls to 0.16 m, still above the 0.15 m minimum, but the area from 0 to 40° drops to 0.087 m·rad and the area from 30 to 40° to 0.027 m·rad. Both fail.
- Make it loll. Raise KG to 8.6 m. GM0 becomes −0.24 m and the ship will not stay upright. It settles at 17.5° to either side, where the wider waterplane gives back stability. The wall-sided formula, tan θ = √(2 × 0.241 / 4.82), gives the same 17.5°. Now add a transverse moment of 500 t·m: the ship lolls only to starboard, at 20.0°.
- Move a weight across. Return KG to 7.6 m and set the transverse moment to 3 000 t·m. The ship lists 11.6°. The small-angle formula, tan θ = w·d / (Δ GM), predicts 13.1°, because it ignores the extra stability the wall sides give. On the low side the area from 0 to 30° falls to 0.052 m·rad and fails.
- Slack tanks. Set the free surface moment to 6 000 t·m. G rises virtually by FSM/Δ = 0.353 m, GM0 falls to 0.41 m and the maximum GZ to 0.374 m. The effect is the same as lifting real weight by 0.35 m.
- Stiff against tender. Set KG to 5 m, heel to 15° and press "Let it roll". The first swing takes 8.7 s and later swings 8.8 s, the value 2πk/√(g GM) gives. Now set KG to 8.2 m. The formula gives 40.2 s, and the first swings take about 33 to 37 s, lengthening as the roll dies down. Release from 2° instead and the swing takes 40.1 s.
- Freeboard and capsize. With KG 7.6 m, set the heel to 50° and let it roll: it comes back. From 52° it passes the angle of vanishing stability, 51.3°, and capsizes. Then cut the depth from 11 to 9 m: the deck edge goes under at 11.3°, stability vanishes at 22.0° and five criteria fail, all with the same GM0.
The model behind it
The hull. The section is a rectangle of beam B and depth D with round bilges of radius r; r = 0 gives a box barge. The ship is treated as prismatic: every section along its length is the same. Section areas then stand for volumes, and the length L matters only for the displacement in tonnes. The deck is closed and watertight, so the section can heel to any angle. Sea water has a density of 1.025 t/m³.
A0 = immersed area upright = B·T − 2 (1 − π/4) r² (for T ≥ r) Δ = 1.025 × L × A0 KB = height of the centroid of A0 BM = I / V = (B_wl³ / 12) / A0 GM0 = KB + BM − KG − FSM / Δ
The inclined waterline. At each heel the simulator finds the waterline that keeps the immersed area equal to A0. In ship axes, with y across and z up from the keel, the immersed part is where the inequality below holds. The offset c is solved by Newton steps, with the change of area per unit offset equal to the waterline length, kept inside a bisection bracket. The immersed polygon comes from clipping the section against that line (Sutherland and Hodgman). Its centroid is B.
immersed: −y sin θ + z cos θ ≤ c, with c chosen so that area = A0 KN(θ) = y_B cos θ + z_B sin θ GZ(θ) = KN(θ) − KG_f sin θ − (w·d / Δ) cos θ, KG_f = KG + FSM / Δ
KN is solved exactly at every degree from 0 to 180° and interpolated between (cubic). The equilibrium, vanishing and deck-edge angles are refined with fresh exact solves. The bilge arcs are drawn with 16 chords each, which trims 0.16 % off the area of each quarter circle. KN is what a real ship's cross curves tabulate; here it comes from a much simpler shape. For the upright properties of a real hull, see hydrostatics basics.
Free surface and weight moments. A slack tank is a virtual rise of G by FSM/Δ, held constant with heel. A real tank's free surface moment changes as it heels. A transverse moment w·d moves G off the centreline by w·d/Δ, which subtracts (w·d/Δ) cos θ from GZ. The dashed curve shows GZ with G back on the centreline.
Checks against the classic formulas. Up to deck-edge immersion and bilge emergence, a section with vertical sides follows the wall-sided formula exactly. The simulator agrees with it to 10⁻⁷ m.
GZ = sin θ (GM + ½ BM tan² θ) wall-sided tan θ_list ≈ w·d / (Δ GM) small angles tan θ_loll = √(−2 GM / BM) wall-sided, GM < 0
IS Code criteria. The six general criteria of the 2008 IS Code, Part A 2.2, are checked on the curve, the same tests as in the GZ curve calculator. Areas are integrated with Simpson's rule and metre-radians. With a transverse moment, they are applied to the net curve on the low side, measured from upright; the Code itself assumes an upright ship, so this reading is conservative. The down-flooding angle θf only caps the 40° limit; the section has no openings.
area 0–30° ≥ 0.055 m·rad area 0–40° (or θf) ≥ 0.090 m·rad area 30–40° (or θf) ≥ 0.030 m·rad GZ at 30° or more ≥ 0.20 m angle of max GZ ≥ 25° GM0 ≥ 0.15 m
Free roll. The roll is one degree of freedom about G in still water, with linear damping. The radius of gyration k includes the water that rolls with the hull; 0.40 B is a common estimate. The damping ratio ζ = 0.05 is an assumed light value; real ships range from about 0.02 to 0.10, more with bilge keels and speed. The equation is integrated with fourth-order Runge–Kutta steps of at most 4 ms. The roll stops when it settles or passes 90°.
Δ k² θ'' + c θ' + Δ g GZ(θ) = 0, c = 2 ζ ω Δ k², ω = √(g GM_eq) / k T0 = 2π k / √(g GM0) small rolls, g = 9.81 m/s² T = 2 C B / √GM0 IS Code 2.3.4 estimate, for comparison
GM_eq is the slope of the GZ curve at the equilibrium, so a listed or lolled ship rolls about its own resting angle. The roll period calculator uses the IS Code estimate, and the initial stability calculator does the GM, list and free surface sums for a real ship's KM.
Limits. No trim, no superstructure, no waves or wind, and no change of shape along the ship. A real ship's GZ curve comes from its cross curves at the actual trim and must be checked against the approved stability booklet. Build a real loading condition with the loading condition calculator.
Sources. IMO Resolution MSC.267(85), International Code on Intact Stability, 2008, Part A 2.2 and 2.3.4. Derrett and Barrass, Ship Stability for Masters and Mates, 7th ed. (2012): transverse statical stability, list, angle of loll, free surface effect and the wall-sided formula. Rawson and Tupper, Basic Ship Theory, 5th ed., ch. 4 (stability) and ch. 11 (rolling). Sutherland and Hodgman, "Reentrant polygon clipping", Communications of the ACM 17 (1974) 32–42.
FAQ
Why can a ship pass the GM test and still fail the criteria?
GM describes only the first few degrees. The areas depend on the whole curve, and freeboard, beam and KG shape it at large angles. A low deck edge or a high G can leave little area beyond 30°.
How do I tell a list from a loll?
A list comes from a weight off the centreline and stays on one side. A loll comes from a negative GM and the ship can flop to either side. The cure for a loll is to lower G, for example by ballasting low tanks one at a time. Moving weight to the high side can make it flop over violently.
Why does a tender ship roll more slowly?
The restoring moment is the displacement times GZ, so a small GM gives a weak spring and a long period. Masters use this the other way round: a short, jerky roll means a large GM.
Why does free surface count as a rise of G?
When the ship heels, the liquid in a slack tank flows to the low side and moves its weight across. That shift reduces GZ by the same amount as raising G by FSM/Δ.
Can I use this to check my ship?
No. It shows the behaviour of a simple prismatic section. A real ship's stability comes from its hydrostatic tables, cross curves and approved stability booklet, and an inclining experiment fixes its lightship KG.