Ship stability explained: GM, the GZ curve and IMO criteria
Transverse stability is the ship's ability to come back upright after wind, waves or a shifted weight have heeled her. Two things describe it. The metacentric height GM governs small angles: whether the ship is stable at all, how far she lists under an off-centre weight and how quickly she rolls. The righting lever curve, GZ against heel angle, governs everything beyond that: how much energy the ship can absorb before she reaches the angle where she floods or capsizes.
A deck officer works with both every time cargo, ballast or bunkers move. This guide explains the points K, B, G and M, shows how GM is built up from the hydrostatic tables and the loading condition, applies the free-surface correction, finds the list from a weight shift, uses the rolling period as a check, and then goes through the intact stability criteria of the IMO 2008 IS Code that every seagoing condition has to meet. The formulas are those of Derrett and Barrass, Ship Stability for Masters and Mates, and of IMO Resolution MSC.267(85).
K, B, G and M
All vertical positions are measured from the keel, K.
- B, the centre of buoyancy, is the centroid of the underwater volume. The buoyancy force acts upwards through it. When the ship heels, the underwater shape changes and B moves towards the low side.
- G, the centre of gravity, is where the whole weight of ship, cargo, fuel and stores acts downwards. It moves only when weights are loaded, discharged or shifted, or when liquid is free to run across a tank.
- M, the transverse metacentre, is the point where the line of buoyancy of the slightly heeled ship crosses the centreline. For small angles it stays in one place, so the ship behaves like a pendulum hung from M.
The height of M comes from the hull form alone: KM = KB + BM, where BM is the transverse second moment of the waterplane area divided by the displaced volume. A wide waterplane gives a large BM, which is why beamy ships are stiff. The metacentric height is the distance from G up to M:
GM = KM − KG
If G is below M, the weight and the buoyancy form a couple that turns the heeled ship back upright. The horizontal distance between the two forces is the righting lever GZ, and at small angles GZ ≈ GM × sin θ. If G is above M the couple capsizes the ship further: GM is negative and the ship will not stay upright. The figure further down heels a section for you and draws B, G and GZ at every angle, and the ship stability simulator adds weight shifts, free surface and rolling.
Where KM and KG come from
KM is read from the ship's hydrostatic tables at the present draft or displacement. It is a property of the hull, so it changes with draft but not with how the ship is loaded. The hydrostatics calculator covers the displacement, TPC and trim quantities that sit in the same tables.
KG comes from the loading condition. Every item on board is listed with its weight and the height of its centre above the keel, starting with the lightship, and the moments are summed:
Δ = Σ w
KG = Σ (w × VCG) ÷ Δ
The loading condition calculator does this sum and interpolates KM, draft and trim from a pasted hydrostatic table. The lightship weight and KG at the top of that list were measured once, by an inclining experiment: known weights are moved across the deck, pendulums record heels of 1° to 4°, and GM follows from the slope of heel against heeling moment. Every KG you calculate afterwards rests on that test.
The free-surface correction
A tank that is neither full nor empty is slack. When the ship heels, the liquid runs to the low side and its centre of gravity moves with it, which reduces the righting moment exactly as if G had risen. The loss is treated as a virtual rise of G, the free-surface correction (FSC). For a rectangular liquid surface of length l and breadth b:
FSM = (l × b³ ÷ 12) × ρ_liquid ÷ n² free surface moment, t·m
FSC = Σ FSM ÷ Δ virtual rise of G, m
GM_fluid = KM − KG − FSC
Here n is the number of equal longitudinal subdivisions of the tank. Three points follow from the formula.
- The breadth is cubed. A wide tank costs far more GM than a narrow one, and one centreline bulkhead (n = 2) cuts the moment to a quarter.
- The quantity of liquid does not appear. A tank with a little liquid in it has the same free-surface moment as one nearly full, as long as the surface is free. Only pressing the tank up or emptying it removes the effect.
- The density is that of the liquid in the tank, not of the water the ship floats in.
For real tanks, which are seldom rectangular, take the free-surface moments from the tank tables in the stability booklet and add them up. The GM that matters for every check below is the fluid GM.
List from a weight shift
A weight w moved a distance d across the ship creates a heeling moment w × d. The ship lists until the righting moment Δ × GZ balances it, and with GZ ≈ GM × sin θ at small angles this gives:
tan θ = (w × d) ÷ (Δ × GM_fluid)
A weight shifted vertically also moves G by w × d_vertical ÷ Δ. A weight loaded or discharged changes both the displacement and KG, by taking moments about the keel:
Load: KG′ = (Δ × KG + w × Kg) ÷ (Δ + w)
Discharge: KG′ = (Δ × KG − w × Kg) ÷ (Δ − w)
with Kg the height of the weight above the keel. In both cases the free-surface correction is recalculated with the new displacement. KM is held constant for a small weight; if the weight is more than about 2 % of the displacement, read KM again at the new draft.
The list formula holds up to about 10°. Beyond that the lever is no longer GM × sin θ and the angle has to be read from the GZ curve.
Worked example: GM with free surface, then list
A ship displaces 24 000 t. The hydrostatic tables give KM = 9.40 m and the loading condition gives a solid KG of 8.10 m. One sea-water ballast tank (1.025 t/m³), 18 m long and 12 m wide with no centreline division, is slack. Breadth is 28 m, waterline length 165 m and mean draft 9.5 m. A 150 t weight is then shifted 12 m to starboard at the same height.
- Solid GM: 9.40 − 8.10 = 1.30 m.
- Free surface moment: 18 × 12³ ÷ 12 = 2 592 m⁴, times 1.025 = 2 656.8 t·m.
- Free-surface correction: 2 656.8 ÷ 24 000 = 0.111 m.
- Fluid GM: 1.30 − 0.111 = 1.189 m. That is above the IMO minimum of 0.15 m.
- Heeling moment: 150 × 12 = 1 800 t·m.
- List: tan θ = 1 800 ÷ (24 000 × 1.189) = 0.0631, so θ = 3.61° to starboard.
Without the slack tank the same shift would give 3.30°. If the tank had a centreline bulkhead, the moment would fall to 664.2 t·m, the correction to 0.028 m, the fluid GM would be 1.272 m and the list 3.37°. The initial stability calculator reproduces all of these figures: choose "From one rectangular slack tank" and enter the values above.
Rolling period as a check on GM
A ship with a large GM snaps back quickly; one with a small GM rolls slowly. The 2008 IS Code gives an approximate natural rolling period for use when nothing better is available:
T = 2 × C × B ÷ √GM
C = 0.373 + 0.023 × (B ÷ d) − 0.043 × (L_wl ÷ 100)
with B, d and L_wl in metres, GM the fluid GM in metres and T in seconds for a full roll, out and back. For the example ship, C = 0.373 + 0.023 × 2.947 − 0.043 × 1.65 = 0.370 and T = 2 × 0.370 × 28 ÷ √1.189 = 19.0 s.
The formula turns round: GM = (2 × C × B ÷ T)². Time ten full rolls in calm water or a gentle beam swell and divide by ten. If the example ship is observed rolling at 21.5 s instead of the 19.0 s predicted, the implied GM is 0.93 m, about a quarter of a metre less than calculated. That is the signal to look for a slack tank nobody counted, a wrong weight or a wrong centre in the loading condition. The roll period calculator works in both directions.
Treat the result as an estimate. The coefficient C stands for a roll radius of gyration of roughly 0.35 to 0.40 B, and ships with an unusual weight distribution can be 10 to 15 % away from it.
Why GM is not enough: the GZ curve
GM is the slope of the GZ curve at the origin. It says nothing about what happens once the deck edge goes under, the bilge comes out, or water reaches an opening. Two ships with the same GM can have very different curves: a high-freeboard ship keeps gaining lever to a large angle, while a low-freeboard ship with the same initial slope peaks early and loses her stability soon after.
The GZ curve of a loading condition comes from the cross curves (KN values) in the stability booklet, corrected for the actual fluid KG:
GZ = KN − KG × sin θ
Four things are read from it: the maximum lever and the angle where it occurs, the range over which GZ stays positive, the down-flooding angle θf at which openings that cannot be closed weathertight go under, and the area under the curve. The area, in metre-radians, is the work needed to heel the ship, so it measures how much wind and wave energy she can take.
In the figure below, move the heel angle and watch B travel to the low side while G stays put. The horizontal gap between the two forces is GZ, and the dot on the curve is the same number. Past the angle where the deck edge goes under, the waterplane stops widening, B travels less, and the curve turns down. Lower the deck and the curve peaks earlier; raise KG and the whole curve drops.
GM 0.76 m, maximum GZ 0.55 m at 30°, stability vanishes at 51°.
| Criterion | Required | This condition | Result |
|---|---|---|---|
| Area 0–30° | ≥ 0.055 m·rad | 0.140 m·rad | Pass |
| Area 0–40° | ≥ 0.090 m·rad | 0.228 m·rad | Pass |
| Area 30–40° | ≥ 0.030 m·rad | 0.087 m·rad | Pass |
| GZ at 30° or more | ≥ 0.20 m | 0.55 m | Pass |
| Angle of maximum GZ | ≥ 25 ° | 30° | Pass |
| GM0 | ≥ 0.15 m | 0.76 m | Pass |
The hull is 20 m wide at a draught of 7 m, with a 2 m bilge radius and no free surface. The immersed shape, B and GZ at each angle are computed by the model behind the ship stability simulator, which is checked against hand calculations before every release. Raise KG to see GM go negative and the ship loll; lower the deck to see a low-freeboard curve peak early and fall away. Without JavaScript the figure shows the ship at 25°.
The IMO intact stability criteria
Part A of the 2008 IS Code sets general criteria on the GZ curve, with free surface already included. The GZ curve calculator checks them from a curve pasted out of the loading computer.
| Criterion (IS Code Part A 2.2) | Required | Calculator's default curve |
|---|---|---|
| Area under GZ, 0° to 30° | ≥ 0.055 m·rad | 0.152 |
| Area, 0° to 40° (or to θf if less) | ≥ 0.090 m·rad | 0.268 |
| Area, 30° to 40° (or to θf if less) | ≥ 0.030 m·rad | 0.116 |
| GZ at an angle of 30° or more | ≥ 0.20 m | 0.69 m |
| Angle of maximum GZ | ≥ 25° | 40° |
| Initial GM, corrected for free surface | ≥ 0.15 m | 1.00 m |
The right-hand column is the calculator's built-in example: a 20 000 t ship with a GM of 1.0 m whose curve peaks at 0.69 m at 40°.
The second part is the severe wind and rolling (weather) criterion of Part A 2.3. A steady beam wind of 504 Pa on the windage area A, acting at a lever Z above half the draft, gives a heeling lever lw1 = P × A × Z ÷ (1000 × g × Δ). The ship heels to θ0 under it, which must not exceed 16° or 80 % of the deck-edge immersion angle. She is then assumed to roll to windward by an angle θ1, taken from the Code's tables for breadth to draft ratio, block coefficient, bilge keels and rolling period, and to be hit by a gust lever lw2 = 1.5 × lw1. The area under GZ above lw2 on the leeward side (area b, counted up to the least of θf, 50° and the second intercept) must be at least the area a between the curves from the windward roll angle to the first intercept. For the default example lw1 = 0.065 m, θ0 = 3.8°, θ1 = 16.2°, area a = 0.049 and area b = 0.301 m·rad, so it passes.
Ships carrying grain have a further check under the International Grain Code, which the same calculator applies: heel from the assumed grain shift no more than 12°, residual area at least 0.075 m·rad and GM at least 0.30 m.
Stiff and tender ships
The minimum GM is 0.15 m, but working values are usually between about 0.5 and 3 m depending on the ship. More is not simply better.
- A stiff ship has a large GM and a short rolling period. She returns violently, which strains lashings, shifts cargo and wears out the crew. As a rule of thumb the calculator calls a period under 10 s stiff.
- A tender ship has a small GM and a long, easy roll. She is comfortable but has little margin: a slack tank, absorbed water in deck cargo or icing can take what GM is left. A period over 20 s is flagged as tender.
These bands are a comfort and cargo-securing guide, not a regulation. The rolling period also matters in a seaway: when the period at which the ship meets the waves comes close to her natural period, rolling builds up, and a change of course or speed is needed to move away from it.
Common mistakes
- Using the solid GM. Every criterion and the list formula need the fluid GM. Forgetting the tanks that are being worked during cargo operations is the classic error.
- Taking the breadth of the ship instead of the tank, or forgetting to divide by n² for a subdivided tank. Because of the cube, either slip changes the answer several times over.
- Assuming a nearly empty or nearly full tank has no free surface. It has the full moment until it is pressed up or dry.
- Reading KM at the wrong draft, or keeping the old KM after loading a large weight.
- Using the list formula at large angles. Past about 10° read the angle off the GZ curve.
- Treating loll as list. A ship with negative GM flops to an angle of loll on either side. Moving weight to the high side makes her fall across to a larger angle on the other. The cure is to lower G, by filling low tanks and removing free surface, never by moving weight to the high side.
- Pasting a GZ curve that does not include free surface, or one that stops short of 40° or the flooding angle, into the criteria check.
- Mixing units in the areas. The criteria are in metre-radians, so angles must be converted from degrees before integrating.
Limits of the method
GM-based results are small-angle results. They assume the heeling moment is applied slowly, that KM stays where the tables put it, and that tank surfaces are rectangular unless you use the tank-table moments. Wind, grain shift and icing moments are separate calculations.
The general criteria above are not the whole Code. Passenger ships, ships with timber deck cargo, large container ships, fishing vessels and offshore vessels have additional or alternative criteria that the calculators do not check. Damage is another subject again: a flooded space changes the waterplane, KB and BM, as the bilged compartment calculator shows for a box-shaped hull, and damage stability rules assess the whole curve after flooding.
Two situations reduce GM without any change in the loading condition. In dry dock, the upthrust on the keel blocks acts like a weight removed from the keel, and the GM at the critical instant has to be checked with the dry-dock critical instant calculator; grounding at one end has the same effect. And any longitudinal shift that changes the draft, which you can follow with the trim change calculator, also changes KM.
The approved stability booklet and the ship's loading computer are the governing documents for any real condition. Hand calculations and these calculators are an independent check on them, not a replacement.
For more practice, the maritime formula sheet lists these relations in one place and the maritime practice problems let you test yourself.
Calculators used in this guide
- MaritimeInitial stability GM and listGM from KM and KG with free-surface correction, list from a weight shift, loading or discharge, rolling period and the IMO minimum-GM check.
- MaritimeGZ curve and intact stability criteriaIS Code 2008 general criteria and the severe wind and rolling (weather) criterion from a pasted GZ curve, with the grain heeling check.
- MaritimeLoading condition from an item listDisplacement, deadweight, KG with free surface, GM, trim, end drafts and list from a pasted list of weights and centres and a hydrostatic table.
- MaritimeInclining experimentGM and lightship KG from an inclining test: weight-movement moments against pendulum tangents by least squares, free surface and the lightweight survey.
- MaritimeRoll period and synchronous rollingNatural roll period from GM and beam (IS Code) or GM from an observed period, encounter period for speed and heading, synchronous and parametric roll checks.
- MaritimeHydrostatics basicsDisplacement from draft and block coefficient, TPC, fresh and dock water allowance, trim and end drafts from LCG, LCB and MCTC.
- MaritimeDry-docking critical instantUpthrust on the keel blocks and the virtual loss of GM at the critical instant of dry-docking, the draft then, and the largest safe trim.
- MaritimeBilged compartment (lost buoyancy)Sinkage, end drafts, trim, freeboard and the new GM of a box-shaped vessel after a compartment is bilged, by the lost buoyancy method.
Results are estimates. They do not replace the governing code, class or flag requirements, or a check by a competent person. Report an error in this guide
More guides
- Cargo securing calculation to CSS Code Annex 13
- How to do a draft survey calculation, step by step
- How EU ETS and FuelEU Maritime costs are calculated
- How to calculate a ship's CII rating, step by step
- Hull and propeller performance in service with ISO 19030
- Ship resistance and powering: from hull to installed power