Free Surface Effect: FSM, GM Loss and Slack Tanks

Free surface effect cuts GM when a tank is part full. See the FSM formula, the subdivision rule, slack versus pressed-up tanks and how the IS Code applies it.

Contents17 sections

Free surface effect is the loss of effective metacentric height (GM) caused by liquid in a partly filled tank. When the ship heels, the liquid surface stays level and liquid moves to the low side. The ship behaves as if its center of gravity G had risen by the free surface correction GG′.

The 2008 IS Code, resolution MSC.267(85), gives the free surface moment of a vertical-sided tank as l b³ ρ / 12 (Annex 1, 2.1.1). The correction is that moment divided by displacement. Part A, 2.1.2 of the Code makes it mandatory to account for free surface in all loading conditions. The method is in Part B, 3.1, which is recommendatory.

What free surface effect is

Free surface effect is a reduction of the righting ability of a ship, caused by liquid that moves inside a tank when the ship heels. A solid weight stays where it is stowed. A liquid with a free surface does not. The metacentric height of the ship, GM, is measured with all weights fixed, so the moving liquid has to be added as a correction.

Liquid shift and the virtual rise of G

A slack tank is a tank that is neither empty nor full. Figure 1 shows a slack tank in a ship heeled by 6 degrees.

levelgg₁θ = 6.0°Tank 16 m wide, 3 m deep, half full
Figure 1. In a heeled ship the liquid surface in a slack tank stays level. The wedge of liquid moves to the low side, so the centroid shifts from g to g₁. The shift acts as a rise of G. The tank is an example.Move the heel. Up to 10.6 degrees the exact loss of GZ and GG′ sin θ agree within 2 per cent. Past that angle the surface meets the tank top and bottom, the liquid can move no further, and the small-angle form overstates the loss.

The surface of the liquid stays level with the sea surface. A wedge of liquid leaves the high side and collects on the low side. The center of the liquid, g, moves to g1, and the center of gravity of the whole ship moves with it toward the low side.

That transverse shift reduces the righting lever GZ. For small angles, the reduction equals the reduction that a rise of G would give. The ship therefore behaves as if G were at a virtual position Gv, above the solid position. The distance GGv is the free surface correction, written GG′ in this article.

In the example of Figure 1, the liquid centroid moves 1.49 m across at 6 degrees. The loss of GZ is 0.060 m by the exact geometry and 0.059 m by the small-angle form, a difference of 1 per cent.

Solid GM and fluid GM

Solid GM is the metacentric height with all liquids treated as frozen; fluid GM is the solid GM less the free surface correction. Fluid GM is the value the stability criteria use. Part A, 2.3 of the 2008 IS Code defines GM in the weather criterion as the metacentric height “corrected for free surface effect”.

The correction does not move any weight upward. The liquid stays at the bottom of the tank. The rise of G expresses a transverse shift as a vertical one. One number then corrects both GM and the GZ curve.

Free surface moment and free surface correction

The free surface moment of a vertical-sided tank is l b³ ρ / 12 in tonne-meters, and the correction is that moment over displacement. The 2008 IS Code prints both expressions in Annex 1, paragraph 2.1.1, in its guidance for the inclining experiment.

IS Code 2008, Annex 1, 2.1.1

Free surface moment

$$\textcolor{#eb6834}{M_{\mathrm{fs}}} = \frac{l\,\textcolor{#2a78d6}{b}^{3}\,\textcolor{#1baf7a}{\rho_{t}}}{12} = \textcolor{#1baf7a}{\rho_{t}}\,i$$

where

\(M_{\mathrm{fs}}\)free surface moment of one tank with parallel vertical sidest m
\(l\)length of the tankm
\(b\)breadth of the tank, the breadth of the liquid surfacem
\(\rho_{t}\)specific gravity of the liquid in the tankt/m³
\(i\)second moment of area of the liquid surface about its fore-and-aft centroidal axis, l b³ / 12m⁴

The term l b³ / 12 is the second moment of area i of the liquid surface about its own fore-and-aft centroidal axis. Multiplying i by the density of the liquid gives the moment in tonne-meters. The Code writes the density as ρt, the specific gravity of the liquid in the tank, in tonnes per cubic meter.

IS Code 2008, Annex 1, 2.1.1; Part A, 2.2.4

Free surface correction and fluid GM

$$\textcolor{#eb6834}{\mathit{GG'}} = \frac{\sum M_{\mathrm{fs}}}{\textcolor{#2a78d6}{\Delta}}$$
$$\textcolor{#1baf7a}{\mathit{GM}_{\mathrm{f}}} = \mathit{KM} - \mathit{KG} - \textcolor{#eb6834}{\mathit{GG'}}$$

where

\(\mathit{GG'}\)free surface correction, the virtual rise of the center of gravitym
\(\sum M_{\mathrm{fs}}\)sum of the free surface moments of the tanks taken into accountt m
\(\Delta\)displacement of the ship in the loading conditiont
\(\mathit{GM}_{\mathrm{f}}\)metacentric height corrected for free surface effect, not less than 0.15 m under Part A, 2.2.4m
\(\mathit{KM}, \mathit{KG}\)height of the metacenter and of the center of gravity above the keelm

Every slack tank adds its own moment. For each loading condition, the moments of the tanks taken into account are summed and divided by the displacement in that condition.

Second moment of area and the cube of the breadth

The breadth of the surface enters as its cube, so breadth is the controlling dimension of every slack tank. Annex 1, 2.1.1 of the 2008 IS Code states that the moment “increases by the third power” as the width of the tank increases. It also states that “the distance available for the liquid to shift is the predominant factor.”

Figure 2 shows the effect in the example ship used for every figure on this page. The ship is a box 100 m long and 20 m wide at a draft of 6 m. Its displacement is 12,300 t, and its solid GM is 1.56 m.

WLb = 16.0 mMGGv0.000.250.500.751.000369121518Breadth b (m)GG′ (m)
Figure 2. The free surface moment of a vertical-sided tank grows with the cube of the breadth, so the rise of G, shown as GG′, climbs steeply as the tank widens. The ship and the tank are an example, not a real ship.Move the breadth. Doubling the breadth multiplies the free surface correction by eight. The depth of liquid, the height of the tank and its position across the ship do not change it.

A sea-water ballast tank 20 m long and 16 m wide has an i of 6,827 m⁴. Its free surface moment is 6,997 t m. Divided by 12,300 t, the moment gives GG′ of 0.57 m. The fluid GM falls from 1.56 m to 0.99 m. A tank 8 m wide gives one-eighth of that correction, 0.07 m.

The same paragraph of the Code states that the correction is “independent of the height of the tank in the ship, location of the tank, and direction of heel.” A double-bottom tank and a deep tank of the same plan dimensions give the same correction. A wing tank and a centerline tank of the same breadth also give the same correction.

Density, displacement and units

The free surface moment scales with the density of the liquid in the tank, and the correction scales inversely with the displacement of the ship. A tank of fuel oil gives a smaller moment than the same tank of sea water. A heavier cargo liquid gives a larger one.

The same tank has a larger effect on a light ship than on a loaded ship, because the displacement is smaller. Part B, 3.1.12 of the 2008 IS Code reflects this: its small-tank test uses the minimum displacement of the ship, Δmin.

The units are consistent. The moment is in t m, the displacement is in t, and the correction is in meters. A tank table that gives the second moment i in m⁴ serves any liquid: the moment is ρ i for the density actually carried.

Tanks with sloped or curved sides

For a tank whose sides are not vertical, the free surface moment changes with the filling level, because the breadth of the surface changes. A double-bottom tank with a sloped hopper side has a narrow surface near the bottom and a wide surface higher up. A tank in the bilge or near the ends of the ship changes shape the same way.

The second moment of area depends on the plan shape of the surface, not on the cross-section of the tank. A designer calculates i for the actual plan of the surface at each filling. Part B, 3.6.4.6 of the 2008 IS Code recommends that the tank tables of the booklet carry free surface data for each tank. The hydrostatics and Bonjean curves article covers the related tables of the hull.

Tank breadth and subdivision

A longitudinal bulkhead reduces free surface effect because it divides the breadth; a transverse bulkhead does not, because it divides only the length. Figure 3 compares four arrangements of the example tank in plan view.

UndividedFSM 100 %Two acrossFSM 25 %Three acrossFSM 11 %Two alongFSM 100 %Plan view of the example tank: 16 m across the ship, 20 m along it. Bow at the top.A bulkhead along the ship divides the breadth; a bulkhead across the ship does not.
Figure 3. A longitudinal bulkhead divides the breadth of the surface, so each part carries less free surface moment. A transverse bulkhead divides only the length, so the total moment does not change. The tank is an example.

Longitudinal bulkheads

n equal compartments across the ship give a total free surface moment of 1/n² of the undivided tank. Each compartment is b / n wide. Its moment is proportional to (b / n)³, and there are n of them.

Hydrostatic relation

Equal longitudinal subdivision

$$\textcolor{#eb6834}{i_{n}} = \textcolor{#2a78d6}{n}\,\frac{l}{12}\left(\frac{b}{\textcolor{#2a78d6}{n}}\right)^{3} = \frac{\textcolor{#1baf7a}{i_{1}}}{\textcolor{#2a78d6}{n}^{2}}$$

where

\(i_{n}\)total second moment of area of the liquid surfaces after the subdivisionm⁴
\(n\)number of equal compartments across the ship, made by n minus 1 longitudinal bulkheads
\(i_{1}\)second moment of area of the undivided tank, l b³ / 12m⁴
\(l, b\)length and breadth of the undivided tankm

One centerline bulkhead gives two compartments and one quarter of the moment. In the example ship the correction falls from 0.57 m to 0.14 m. Two bulkheads give three compartments and one ninth of the moment, 0.06 m.

Figure 4 shows the undivided tank and the divided tank in the same ship under the same wind heeling moment of 2,000 t m. The steady heel is 8.9 degrees with the undivided tank and 6.4 degrees with the divided tank.

One tank 16 m wideGG′ 0.57 m, fluid GM 0.99 mwindGZ0.3 m−0.3 m10° port10° stbd0heel 8.9° to starboardDivided on the centerlineGG′ 0.14 m, fluid GM 1.41 mwindGZ0.3 m−0.3 m10° port10° stbd0heel 6.3° to starboardshift of the liquid center, g to g1; a ring when under 0.6 mGZ righting lever with the liquid movingrighting lever with the liquid frozenwind heeling lever, 2,000 t·m ÷ 12,300 t
Figure 4. A centerline bulkhead halves each liquid surface. In the undivided tank the liquid center moves far to the low side and the GZ curve falls. In each half it moves a quarter as far, so the same wind heels the ship less.Two ships take the same wind, from port then starboard. The liquid in the undivided tank moves farther to the low side, so that ship heels and rolls more. Each curve marker settles where the curve meets the wind lever. Four times real time.

The same arithmetic applies to a wide double-bottom tank on a bulk carrier or an oil tanker. Part B, 3.8.2 of the 2008 IS Code recommends an operation manual for a double hull oil tanker of single cargo tank across design. The manual gives the free surface correction of the cargo and ballast tanks during loading, unloading and tank washing.

Transverse bulkheads

A transverse bulkhead divides the tank into a forward part and an after part. Each part has the full breadth b. The moments of the parts add to the moment of the undivided tank, because the moment is proportional to the length. The fourth plan in Figure 3 shows this case: the moment stays at 100 per cent.

Openings, gaps and swash plates

A division reduces the free surface moment only if the liquid cannot pass from one compartment to the next as the ship heels. Two compartments joined by an opening at the bottom act as one wide surface. The same applies to two tanks joined by an open cross-connection.

The capsize of the stern trawler Fraoch Ban off Shetland on 15 August 1999 shows the mechanism with a bulk catch. MAIB Report 12a/2000 found that “the arrangement of the pound boards in the hold did not limit the free surface sufficiently.” The report adds that the divisions “did not break up the free surface satisfactorily, because the catch could move from one section to another through the gaps under the pound boards.”

Filling level: slack, nominally full and pressed up

The free surface correction follows the breadth of the liquid surface, not the quantity of liquid in the tank. In a tank with vertical sides, a thin layer gives the full correction. Annex 1, 2.1.1 of the 2008 IS Code states that “even the smallest amount of liquid in the bottom of a wide tank or bilge is normally unacceptable” before an inclining experiment.

Figure 5 shows the correction against filling for a tank with sloped lower sides.

surface 14.3 m wideTank section, 3 m deep0.00.20.40.60255075100Filling (per cent of capacity)GG′ in the correction (m)98 %
Figure 5. The free surface correction follows the breadth of the liquid surface, not the quantity of liquid. The example tank has sloped lower sides, so the surface widens with depth and the correction rises, then stays level. The tank is an example.Move the filling. A layer 0.21 m deep, 5 per cent of capacity, already gives a surface 10.8 m wide and a correction of 0.18 m. The correction stays at 0.57 m from 1.5 m depth up to 98 per cent full.

In the example tank, a layer 0.21 m deep holds 5 per cent of capacity. Its surface is already 10.8 m wide, and the correction is 0.18 m. From 1.5 m depth the sides are vertical, the surface is 16 m wide, and the correction stays at 0.57 m.

Part B, 3.1.2 of the 2008 IS Code sets the limits. Free surface effects should be considered whenever a tank is filled to less than 98 per cent of full. A tank filled to 98 per cent or more is “nominally full”, and its free surface effect need not be considered.

Cargo tanks are the exception in the same paragraph. Nominally full cargo tanks should be corrected at 98 per cent filling. The GM correction is then based on the moment of inertia of the surface at 5 degrees of heel, divided by displacement.

The inclining experiment uses a stricter definition. Annex 1, 2.1.3 of the Code defines pressed up as “completely full with no voids caused by trim or inadequate venting.” It adds that anything less than 100 per cent full, “for example the 98% condition regarded as full for operational purposes, is not acceptable.”

Tanks at a fixed level and tanks at a variable level

Part B, 3.1.3 of the 2008 IS Code puts every tank in one of two categories, and the category decides which filling level is used. The text reads:

CategoryExamples in Part B, 3.1.3Filling level for the correction
.1 Filling level fixedLiquid cargo, water ballastThe actual filling level used in each tank
.2 Filling level variableFuel oil, diesel oil, fresh water; cargo and ballast during transferThe maximum value attainable between the filling limits envisaged

A ballast tank at a fixed level in the loading condition is in category .1. The same tank moves into category .2 when it is filled or emptied at sea.

Consumables: the tanks assumed slack

For consumables, Part B, 3.1.4 of the 2008 IS Code assumes one transverse pair or one centerline tank of each liquid type is slack. The tanks taken into account are those with the greatest free surface effect. Take a ship with heavy fuel oil, marine gas oil, lubricating oil and fresh water. Every condition in its booklet carries at least four such assumptions.

Consumption changes the filling of these tanks through the voyage. The stability booklet and loading computer article covers how the conditions are built from departure to arrival.

Small tanks and the usual remainders

Part B, 3.1.12 of the 2008 IS Code leaves out a small tank whose moment at 30 degrees is small against the minimum displacement. The test is in the card below.

IS Code 2008, Part B, 3.1.12

Small tanks left out of the correction

$$\frac{\textcolor{#eb6834}{M_{\mathrm{fs}}}}{\textcolor{#2a78d6}{\Delta_{\mathrm{min}}}} < 0.01\ \mathrm{m} \quad \text{at } 30^{\circ}$$

where

\(M_{\mathrm{fs}}\)free surface moment of the tank at an inclination of 30 degreest m
\(\Delta_{\mathrm{min}}\)minimum ship displacement, calculated at the draft d_mint
\(d_{\mathrm{min}}\)minimum mean service draft without cargo, with 10% stores and minimum water ballast, if requiredm

The minimum draft dmin is that of the ship without cargo, with 10 per cent stores and minimum water ballast. If Δmin were 12,300 t, the limit of 0.01 m would correspond to a moment of 123 t m. A lighter Δmin gives a lower limit.

Part B, 3.1.13 adds a second exemption. The usual remainder of liquids in empty tanks need not be taken into account, “provided that the total of such residual liquids does not constitute a significant free surface effect.”

Free surface correction on the GZ curve

Part B of the 2008 IS Code recommends separate corrections to GM and to the GZ curve. A correction calculated at 0 degrees does not hold at every angle of heel. Part B, 3.1.7 states that the two corrections “should be addressed separately”. Part B, 3.1.8 bases the GM correction on the transverse moments of inertia of the tanks at 0 degrees.

The simplest correction of the curve deducts GG′ sin θ at each angle. That is the small-angle form below.

Hydrostatic relation; methods in Part B, 3.1.9

Fluid righting lever, small-angle form

$$\textcolor{#1baf7a}{\mathit{GZ}_{\mathrm{f}}} = \textcolor{#2a78d6}{\mathit{GZ}} - \textcolor{#eb6834}{\mathit{GG'}}\sin\theta$$

where

\(\mathit{GZ}_{\mathrm{f}}\)righting lever corrected for free surface effectm
\(\mathit{GZ}\)righting lever with all liquids treated as solidm
\(\mathit{GG'}\)free surface correction from the 0 degree moments of inertiam
\(\theta\)angle of heel°

Figure 6 shows the corrected curve of the example ship, with GG′ of 0.57 m.

-1.0-0.50.00.51.01.52.02.50102030405060708090Angle of heel (degrees)GZ (m)solidcorrected
Figure 6. The free surface correction of the example tank in Figure 2 lowers the GZ curve of the example ship by about GG′ sin θ at each angle. GM and the area under the curve both fall. The ship is an example.

The correction lowers GM from 1.56 m to 0.99 m. The area under the curve to 40 degrees falls from 0.528 to 0.394 m-rad. Both values stay above the Part A, 2.2 minimums of 0.15 m and 0.090 m-rad, which the intact stability article lists with the other criteria.

The two methods of Part B, 3.1.9

Part B, 3.1.9 of the 2008 IS Code allows two methods for the GZ curve, each “subject to the agreement of the Administration”. The methods are:

  1. A correction based on the actual moment of fluid transfer, calculated for each angle of heel.
  2. A correction based on the moment of inertia at 0 degrees, modified at each angle of heel.

Part B, 3.1.11 recommends that only the selected method is presented in the stability booklet. Where the booklet also describes an alternative for manually calculated conditions, it should explain the differences and give an example correction for each method.

The Code does not use the names “moment method” or “shift method”. Its own terms are “actual moment of fluid transfer” and “moment of inertia, calculated at 0 deg … modified at each angle of heel”.

Where the small-angle form stops working

The small-angle form stops working when the liquid surface reaches the tank top or the tank bottom. From that angle the liquid cannot move farther across. The exact loss of GZ then grows more slowly than GG′ sin θ.

The live version of Figure 1 shows this. The example tank is 16 m wide and half full to 1.5 m. The surface meets the top and bottom at 10.6 degrees, where tan θ equals 1.5 / 8. Up to that angle the two values agree within 2 per cent. At 20 degrees the exact loss is 0.144 m against 0.195 m by the small-angle form.

Part B, 3.1.2 therefore bases the GM correction of a nominally full cargo tank on the inertia at 5 degrees. A surface at 98 per cent meets the tank top at a small angle. The wall-sided formula and the cross curves of stability cover the large-angle geometry of the hull itself.

Status in the IS Code and other instruments

Free surface accounting is mandatory under Part A of the 2008 IS Code, and the method of calculation is recommendatory under Part B. The Code was adopted by resolution MSC.267(85) on 4 December 2008. Part A took effect on 1 July 2010 through the SOLAS amendments of MSC.269(85) and the Load Line Protocol amendments of MSC.270(85).

Part A and Part B of the 2008 IS Code

The 2008 IS Code has two parts and two annexes. Part A is mandatory under SOLAS regulation II-1/2.27. Part B is recommendatory. Annex 1 is guidance for the inclining test, and Annex 2 gives recommendations for skippers of fishing vessels on ice formation.

SOLAS regulation II-1/5.1 applies Part A to ships of 24 m in length and upwards constructed on or after 1 July 2010. Part A, 2.1.2 states that “free surface effects (part B, 3.1) shall be accounted for in all conditions of loading”. The SOLAS Convention article covers how the Code is made mandatory.

Criteria that use the corrected GM

Part A, 2.2.4 of the 2008 IS Code requires an initial GM0 of not less than 0.15 m, and the value is the corrected one. Part A, 2.1.2 makes the correction mandatory in all conditions of loading. The weather criterion of Part A, 2.3 uses the same corrected GM.

The weather criterion compares two areas under the GZ curve. Part A, 2.3.1.4 states that “area b shall be equal to or greater than area a”. Part A, 2.3 contains no factor of 1.4. The weather criterion section of the intact stability article covers the full method.

Ship-type provisions in Part B

Part B of the 2008 IS Code returns to free surface in several ship-type provisions:

  • 3.7.1.3: for ships carrying timber deck cargo, the free surface effect of liquid in tanks is among the factors in the stability calculation.
  • 3.8.2: single cargo tank across double hull oil tankers carry an operation manual with free surface corrections during cargo and ballast operations.
  • 3.8.3: the booklet of a ro-ro passenger ship explains the rapid loss of stability when water enters the vehicle deck.
  • 5.1.5: the number of slack tanks should be kept to a minimum, and the effect of filled pool tanks should be considered.
  • 5.1.6: slack tanks can, “in exceptional cases”, be used to reduce an excessive GM, with “due consideration” to sloshing.

Grain Code

The International Grain Code, resolution MSC.23(59), requires an initial GM of at least 0.30 m after the free surface correction. The requirement is in paragraph A 7.1.3. The Code was adopted on 23 May 1991 and took effect on 1 January 1994.

Grain is not a liquid, but it shifts when the ship heels. The Grain Code treats that shift as an assumed volumetric heeling moment, divided by the stowage factor and the displacement to give a heeling arm. The free surface correction of the tanks and the grain heeling arm are separate terms of the same check.

Stability booklet, loading computer and inclining test

The stability booklet carries the free surface data of each tank, and the loading computer applies it to every loading condition. Part B, 3.6.4.6 of the 2008 IS Code lists “tank sounding tables showing capacities, centres of gravity, and free surface data for each tank” among the booklet contents.

IACS Unified Requirement L5 (Rev.4, June 2020) covers computer software for onboard stability calculations. Part B, 4.1.1.2 of the 2008 IS Code states that a stability instrument is not a substitute for the approved stability booklet.

The inclining experiment needs the free surface of every slack tank to be known exactly, because the test measures GM directly. Part B, 8.2.2.5 of the 2008 IS Code recommends that “all tanks should be empty and clean, or completely full”, and that the number of slack tanks “should be kept to an absolute minimum”.

Annex 1, 2.1.2 gives the limits for the tanks that remain slack:

  • One port and starboard pair, or one centerline tank, of each listed type. The types are fresh water reserve feed, fuel or diesel storage, fuel or diesel day, lubricating oil, sanitary and potable water.
  • Deep tanks 20 to 80 per cent full, and double-bottom tanks 40 to 60 per cent full.
  • Cross-connections, including those through manifolds, closed.
  • No correction for tanks of viscous liquid, such as cold bunker fuel, unless the tanks are heated.

Annex 1, 2.1.4 adds that “it is generally not sufficient to simply pump tanks until suction is lost.” A tank pumped to the suction still holds a layer across its full breadth. The inclining experiment result fixes the lightship KG, and the lightweight and deadweight of the ship follow from it.

Liquid transfer, ballasting and ballast water exchange

A transfer of liquid between two tanks makes both tanks slack while it runs, so the correction can exceed the correction at either end. Part B, 3.1.5 of the 2008 IS Code therefore uses the most onerous transitory stage.

Transfers at sea and the most onerous stage

Part B, 3.1.5 of the 2008 IS Code covers ballast tanks, “including anti-rolling tanks and anti-heeling tanks”, that are filled or discharged during the voyage. For those tanks the correction should “take account of the most onerous transitory stage”. Part B, 3.1.6 lets a ship in liquid transfer operations use the actual filling level of each tank at each stage.

Annex 2 of the Code states the hazard for fishing vessels in icing conditions. Correcting a list “by pumping fuel or water from one tank to another may reduce stability during the process when both tanks are slack.”

Exchange methods under the G6 Guidelines

Ballast water exchange under the BWM Convention uses three methods, and only the sequential method leaves a tank slack. The 2017 Guidelines for ballast water exchange (G6), resolution MEPC.288(71) of 7 July 2017, list them in paragraph 4.4. Figure 7 compares the sequential and flow-through methods on one tank of the example ship. In the example, the fluid GM falls from 1.56 m to between 0.99 m and 0.72 m while the tank is slack.

Sequential: emptied, then refilledemptying: the tank is slackfluid GM (m) over one exchange01fluid GM 0.81 m, tank slackFlow-through: kept fullwater pumped in and out at the overflowfluid GM (m) over one exchange01fluid GM 1.56 mballast water in the tankpumpingGM fluid GM over the exchangefluid GM of the ship with the tank full
Figure 7. The same tank is exchanged by two methods. Emptying and refilling leaves the tank slack for most of the exchange, so its free surface lowers the fluid GM. Pumping water through a full tank leaves no free surface. The ship is an example.The left tank empties and refills while the right tank stays full with water pumped through it. The trace under each ship follows the fluid GM. It drops while the tank is slack and recovers when the tank is full or empty.

In the sequential method a tank “is first emptied and then refilled”. In the flow-through method water is pumped in and leaves through the overflow. In the dilution method water enters at the top and leaves at the bottom at the same flow rate, so the level stays constant. MEPC.288(71) revokes the earlier G6 Guidelines of resolution MEPC.124(53).

Paragraph 5.3 of the 2017 G6 Guidelines asks for stability “not less than those values recommended by the Organization or required by the Administration” at all times. Paragraph 5.4.2 lists “free surface effects on stability and sloshing loads in tanks that may be slack at any one time” as a topic for the ballast water management plan.

Regulation B-4.4 of the Convention protects the master’s decision. A ship “shall not be required to comply” with the exchange regulations if the master reasonably decides that exchange “would threaten the safety or stability of the ship”. The ballast water exchange operations article covers the sequence planning.

Cargo and water that act as a free surface

Any cargo or water that moves freely across the ship as it heels gives a free surface effect, in a tank or not. Four cases recur in the instruments: liquefied bulk cargo, firefighting water, water on a vehicle deck, and water in deck wells.

Liquefied bulk cargo

The IMSBC Code describes cargo liquefaction as a liquid slurry of water and fine solids forming above the solid cargo. The slurry gives a free surface effect. Group A cargoes are the cargoes that can liquefy. The IMSBC Group A cargoes article covers the moisture limits that prevent it.

Firefighting water

SOLAS regulation II-2/20.6.1.4, in the revised chapter II-2 adopted by resolution MSC.99(73) on 5 December 2000, addresses water from fixed water-spraying systems. The text refers to “the serious loss of stability which could arise due to large quantities of water accumulating on the deck or decks.” The revised chapter entered into force on 1 July 2002.

Above the bulkhead deck of a passenger ship, scuppers discharge the water directly overboard. Below it, the Administration can require additional drainage. That system removes not less than 125 per cent of the spraying pumps and hose nozzles combined, with bilge wells at most 40 m apart.

For cargo ships, “the drainage and pumping arrangements shall be such as to prevent the build-up of free surfaces.” If that is not possible, the added weight and free surface are taken into account “to the extent deemed necessary by the Administration” in its approval of the stability information. The SOLAS chapter II-2 article covers the fire protection requirements in full.

Water on a ro-ro vehicle deck

A vehicle deck is an undivided space as wide as the ship, so a shallow layer of water on it has the full breadth. Part B, 3.8.3 of the 2008 IS Code recommends that the booklet of a ro-ro passenger ship explains “the rapid loss of stability which may result when water enters the vehicle deck and the fact that capsize can rapidly follow.”

The MS Estonia sank in the Baltic Sea on 28 September 1994 with the loss of 852 lives. She lost her bow visor at about 01:15, and the visor pulled the bow ramp open onto the car deck. The Joint Accident Investigation Commission (JAIC) final report of December 1997 attributed the capsize to “large amounts of water entering the car deck, loss of stability and subsequent flooding of the accommodation decks” (Chapter 21).

The cube law shows the scale. A centerline division of the same deck would cut its second moment of area to one quarter, because each half has half the breadth.

Deck wells and pool tanks

Part B, 5.3.5 of the 2008 IS Code states that “water trapping in deck wells should be avoided.” If the freeing ports cannot drain the well, the speed of the ship should be reduced or the course changed, or both. Part B, 5.1.5 recommends that the effect of filled pool tanks on passenger ships is taken into consideration.

Free surface in damage conditions

SOLAS regulation II-1/7.6 states that “only one breach of the hull and only one free surface need to be assumed” in the flooding calculations. The regulation is in the revised chapter II-1 adopted by resolution MSC.216(82) on 8 December 2006, in force from 1 January 2009. The assumption applies to the probabilistic method of damage stability for passenger ships and for cargo ships in its scope.

The explanatory notes in resolution MSC.429(98) of 9 June 2017 cover intermediate stages of flooding. The notes recommend intermediate-stage calculations “whenever equalization is not instantaneous”, meaning longer than 60 seconds. Each stage has at least two intermediate filling phases in addition to the full phase.

For each phase except the last, the notes assume a constant volume of floodwater at each heeling angle. “Only one free surface needs to be assumed for water in spaces flooded during the current stage.” In the final phase the water level reaches the sea level outside, so the notes allow the lost buoyancy method.

Water in a compartment open to the sea at equilibrium is treated as lost buoyancy. Before equalization, the notes treat the floodwater as a constant volume with one free surface, and the GZ curve keeps the intact displacement. The probabilistic damage stability article covers the s-factor that uses the resulting GZ curves, and the SOLAS chapter II-1 article covers the regulations.

Anti-heeling and anti-roll tanks

An anti-roll tank uses moving water on purpose, and an anti-heeling tank changes its filling during cargo work. Part B, 3.1.5 of the 2008 IS Code names both among the ballast tanks whose most onerous transitory stage should be used.

An anti-heeling system moves water between wing tanks to hold the ship upright during cargo work. Part B, 5.3.7 recommends that wind heel is not compensated by anti-heeling measures. The exception is a ship that has been “proven by calculation to have sufficient stability in worst case conditions”, subject to the approval of the Administration.

A U-tube anti-roll tank lets water swing from side to side against the roll. Figure 8 shows the example ship in a beam sea at its roll period, without and with a tank tuned to that period. In this stated example the steady roll falls from 10.0 to 3.1 degrees.

Without a tankroll 10.0° each wayroll angle (°), to starboard positive10−10roll now 10.0° to starboardWith a U-tube tankroll 3.1° each wayroll angle (°), to starboard positive10−10roll now 3.1° to starboardwater in the U-tube tankroll angletank water level, starboard high positive
Figure 8. The same ship rolls in the same beam sea, without and with a U-tube tank tuned to its roll period. The tank water swings a quarter period behind the roll, so its weight works against each roll. The values are examples.Both ships roll steadily in the same sea at twice real time. The water in the U-tube tank reaches each side a quarter period after the roll does, so its weight acts against the motion. The traces show that lag.

Hermann Frahm of Hamburg patented the arrangement in the United States. US patent 970,368, “Means for damping the rolling motion of ships”, was filed on 4 May 1909 and issued on 13 September 1910. Its first claim describes “a water vessel at each side of the ship, a cross connection joining the bottoms of said vessels.”

The tank works only if the water is free to move, so its free surface stays in the stability calculation. Part A, 2.1.3 of the 2008 IS Code requires the Administration to be satisfied that the criteria are met with anti-rolling devices in operation. The marine stabilizers article compares tanks with fin and gyro systems.

Casualties where an investigation names free surface

Two UK Marine Accident Investigation Branch reports that name free surface effect as a cause of capsize are those on Fraoch Ban and Sarah Jayne. Both were fishing vessels. SOLAS regulation I/3(a)(vi) states that the regulations, “unless expressly provided otherwise”, do not apply to fishing vessels.

VesselDateReportFinding on free surface
Fraoch Ban, stern trawler15 August 1999, off ShetlandMAIB Report 12a/2000“The arrangement of the pound boards in the hold did not limit the free surface sufficiently, and as a result the vessel capsized and foundered.”
Sarah Jayne, 14.94 m fishing vessel11 September 2012, east of Berry HeadMAIB Report 13/2013, June 2013She “had an insufficient reserve of stability to withstand the sudden flooding and its associated free-surface effect.”

Fraoch Ban was fishing for sand eels stowed in bulk. The MAIB found that 25 tonnes of sand eels gave enough free surface to capsize her. Her stability book was not approved for bulk stowage of fish. Sarah Jayne capsized with about 20 tonnes of catch in her fish hold, with the loss of one life.

The Marine Accident Investigation Branch publishes its reports in full. The Estonia finding on the car deck is in the ro-ro section of this article.

Port state control

Port state control records stability defects under standard deficiency codes. The Paris MoU list of deficiency codes of 1 July 2024 includes these codes:

CodeTitle
01326Stability Information Booklet
02103Stability/strength/loading information and instruments
02107Ballast, fuel and other tanks
02134Loading/Ballast condition
19103Subdivision & Stability - IP Code

The Paris MoU article covers that regime, and the port state control article covers the inspection and detention procedure.

Limitations

The formulas on this page are for a tank with parallel vertical sides and a surface clear of the tank top and bottom. A real tank needs the second moment of its actual surface at each filling, from the approved tank tables.

The example ship is a box of constant section. Its figures show each mechanism and do not give values for any real ship. The stability of a real ship is checked against its approved stability booklet, with the method that the booklet presents.

The 2008 IS Code text cited here is the text of MSC.267(85) as adopted. Later amendments to Part B, such as MSC.415(97) on anchor handling, towing and lifting, are covered in the GZ curve article. The SOLAS chapter II-2 text quoted is the text adopted by MSC.99(73).

National rules for fishing vessels and small craft set their own free surface requirements. The casualty table lists only investigations whose reports name free surface as a cause.

Frequently Asked Questions (FAQs)

What is free surface effect?

Free surface effect is the loss of effective metacentric height caused by liquid in a partly filled tank.

When the ship heels, the liquid surface stays level and the liquid moves to the low side. The ship then behaves as if its center of gravity had risen by the free surface correction.

What is the difference between free surface moment and free surface correction?

The free surface moment is in tonne-meters; the free surface correction is that moment divided by displacement, in meters.

The 2008 IS Code, Annex 1, 2.1.1, gives the moment of a vertical-sided tank as l b cubed times the liquid density, divided by 12. The correction is the virtual rise of G that is deducted from GM.

Does the depth of liquid in a tank change the free surface effect?

No, not in a tank with vertical sides.

Annex 1, 2.1.1 of the 2008 IS Code states that the correction is independent of tank height, location and direction of heel. Depth matters only where the tank sides slope, because the breadth of the surface then changes with depth.

Why does the free surface effect depend on the cube of the tank breadth?

The breadth is cubed because the moment is the second moment of area of the liquid surface.

A wider tank lets more liquid move, and it moves farther. Annex 1, 2.1.1 of the 2008 IS Code states that the moment increases by the third power of the width. Doubling the breadth multiplies the correction by eight.

Does a tank filled to 98 per cent need a free surface correction?

Not unless it is a cargo tank.

IS Code 2008, Part B, 3.1.2, treats a tank at 98 per cent or more as nominally full. Its free surface effect need not be considered. Nominally full cargo tanks should still be corrected at 98 per cent. Their GM correction uses the moment of inertia of the surface at 5 degrees of heel.

Is a pressed-up tank the same as a full tank?

For the inclining experiment, no. Annex 1, 2.1.3 of the 2008 IS Code defines pressed up as completely full, with no voids caused by trim or poor venting.

Anything less than 100 per cent, including the 98 per cent condition used in service, is not acceptable for the test.

Which tanks can be left out of the free surface correction?

Small tanks and the usual remainders in empty tanks can be left out.

IS Code 2008, Part B, 3.1.12, excludes a tank whose free surface moment over the minimum displacement is less than 0.01 m at 30 degrees. Part B, 3.1.13, excludes usual remainders if their total does not give a significant free surface effect.

What filling level is used for fuel oil and fresh water tanks?

The maximum value of the correction between the filling limits envisaged for each tank.

IS Code 2008, Part B, 3.1.3.2, puts consumables in the variable-level category. Part B, 3.1.4, assumes one transverse pair or one centerline tank of each liquid type is slack. The tanks with the greatest effect are chosen.

What filling level is used for a ballast tank at a fixed level?

The actual filling level. IS Code 2008, Part B, 3.1.3.1, puts liquid cargo and water ballast at a fixed level in the first category.

The correction is defined for the filling level actually used in each tank. A ballast tank filled or emptied during the voyage moves into the variable category.

Is the free surface correction applied to GM only or also to the GZ curve?

To both, and separately. IS Code 2008, Part B, 3.1.1, recommends that the initial metacentric height and the righting lever curve are both corrected.

Part B, 3.1.8, takes the GM correction from moments of inertia at 0 degrees. Part B, 3.1.9, gives two methods for the GZ curve.

What are the two methods for correcting the GZ curve?

The actual moment of fluid transfer at each angle of heel, or the moment of inertia at 0 degrees modified at each angle.

Both are in IS Code 2008, Part B, 3.1.9, and each needs the agreement of the Administration. Part B, 3.1.11, recommends that only the selected method is presented in the stability booklet.

At what heel angle does the small-angle free surface formula stop working?

It stops working when the liquid surface reaches the tank top or the tank bottom.

From that angle the liquid cannot move farther across, so GG prime times the sine of the heel overstates the loss of GZ. In a tank 16 m wide, half full to 1.5 m, that happens at about 10.6 degrees.

What minimum GM applies after the free surface correction?

0.15 m under Part A of the 2008 IS Code, and 0.30 m for ships carrying grain.

Part A, 2.2.4, sets the initial GM minimum of 0.15 m. Part A, 2.3, uses the GM corrected for free surface effect in the weather criterion. The International Grain Code, A 7.1, sets 0.30 m after correction for liquids in tanks.

How much does a centerline bulkhead reduce free surface effect?

One centerline bulkhead cuts the free surface moment of the tank to one quarter.

Each half is half as wide, and the moment varies with the cube of the breadth. The two halves together therefore give two times one-eighth. Two longitudinal bulkheads making three equal tanks cut the moment to one ninth.

Do transverse bulkheads reduce free surface effect?

No. A transverse bulkhead divides the length of the tank, and the free surface moment is proportional to the length.

The lengths of the parts add up to the original length, so the total moment does not change.

Do swash plates or pound boards with gaps divide the free surface?

Not if liquid or cargo can pass through the gaps as the ship heels.

The trawler Fraoch Ban capsized in 1999 with a bulk catch of sand eels. MAIB Report 12a/2000 found that the catch moved through the gaps under the pound boards. The divisions did not break up the free surface.

Why can ballast water exchange by the sequential method reduce stability?

Because each tank is slack while it is emptied and refilled.

The 2017 G6 Guidelines, MEPC.288(71), describe the sequential method as emptying a tank and then refilling it. Paragraph 5.4.2 lists free surface effects in tanks that can be slack at any one time as a topic for the ballast water management plan.

Can the master decide not to exchange ballast because of stability?

Yes. Regulation B-4.4 of the BWM Convention is quoted in the G6 Guidelines, paragraph 3.3.

A ship shall not be required to exchange ballast if the master reasonably decides that exchange would threaten the safety or stability of the ship.

Is firefighting water a free surface?

Yes. SOLAS regulation II-2/20.6.1.4, in the revised chapter II-2 adopted by MSC.99(73), refers to the serious loss of stability from large quantities of water on deck.

On cargo ships the drainage and pumping arrangements are to prevent the build-up of free surfaces.

How many free surfaces are assumed in a SOLAS damage stability calculation?

One. SOLAS regulation II-1/7.6, in the text of MSC.216(82), states that only one breach of the hull and only one free surface need to be assumed.

The explanatory notes in MSC.429(98) apply the same assumption to water in spaces flooded during the current intermediate stage.

When are intermediate stages of flooding calculated?

When equalization of the flooding is not instantaneous.

The explanatory notes to SOLAS regulation II-1/7-2.2 in MSC.429(98) recommend intermediate-stage calculations where equalization takes more than 60 seconds. Each stage has at least two intermediate filling phases in addition to the full phase for each flooded space.

Do anti-rolling tanks and anti-heeling tanks count in the free surface correction?

Yes. IS Code 2008, Part B, 3.1.5, names anti-rolling and anti-heeling tanks among the water ballast tanks filled or discharged during the voyage.

For those tanks the most onerous transitory stage should be used. Part A, 2.1.3, requires the criteria to be met with anti-rolling devices in operation.

Can an anti-heeling system be used to compensate wind heel?

Not as a rule. IS Code 2008, Part B, 5.3.7, recommends that wind heel is not compensated by anti-heeling measures.

The exception needs approval by the Administration, where calculation shows enough stability in worst cases such as mechanism failure.

Can slack tanks be used to reduce an excessive GM?

Only in exceptional cases. IS Code 2008, Part B, 5.1.6, states that slack tanks can, in exceptional cases, be used to reduce excessive values of metacentric height.

Due consideration should then be given to sloshing effects. Part B, 5.1.5, recommends keeping the number of slack tanks to a minimum.

How should tanks be prepared for an inclining experiment?

Empty and clean, or completely full, with as few slack tanks as possible.

IS Code 2008, Part B, 8.2.2.5, gives that recommendation. Annex 1, 2.1.2 limits slack tanks to one pair or one centerline tank of listed types. Deep tanks should be 20 to 80 per cent full and double-bottom tanks 40 to 60 per cent, with cross-connections closed.

Why are cross-connected tanks a problem in a free surface calculation?

Because two connected tanks act as one wide surface.

Annex 1, 2.1.2 of the 2008 IS Code recommends that communication between tanks is never allowed during the inclining test. Cross-connections, including manifolds, should be closed. Equal levels in a slack pair can be a warning sign of an open cross-connection.

Who invented the U-tube anti-roll tank?

Hermann Frahm of Hamburg patented it.

His US patent 970,368, for means of damping the rolling motion of ships, was issued on 13 September 1910. Its first claim describes a water vessel at each side of the ship, with a cross connection joining their bottoms.

Which casualty reports name free surface effect as a cause?

Two MAIB reports that do are Report 12a/2000 on the trawler Fraoch Ban and Report 13/2013 on the fishing vessel Sarah Jayne.

Fraoch Ban capsized on 15 August 1999 because the pound boards did not limit the free surface of a bulk catch. Sarah Jayne capsized on 11 September 2012 after sudden flooding and its free-surface effect.

What do port state control officers record about stability information?

Paris MoU inspectors record stability defects under codes 01326, 02103, 02107 and 02134.

In the list of 1 July 2024, 01326 is the Stability Information Booklet and 02103 is stability, strength and loading information and instruments. Code 02107 covers ballast, fuel and other tanks, and 02134 the loading and ballast condition.

Does free surface matter on a ship carrying timber deck cargo?

Yes.

IS Code 2008, Part B, 3.7.1.3, lists the free surface effect of liquid in tanks among the factors for a ship carrying timber deck cargo.

Does a swimming pool on a passenger ship count?

Yes. IS Code 2008, Part B, 5.1.5, states that the negative effect on stability of filled pool tanks should be taken into consideration.

The same paragraph recommends keeping slack tanks to a minimum.

Sources

  1. IMO Resolution MSC.267(85). International Code on Intact Stability, 2008 imo.org
  2. IMO Resolution MSC.23(59). International Code for the Safe Carriage of Grain in Bulk imo.org
  3. IMO Resolution MSC.216(82). Amendments to SOLAS 1974, revised chapter II-1 imo.org
  4. IMO Resolution MSC.429(98). Revised explanatory notes to the SOLAS chapter II-1 subdivision and damage stability regulations imo.org
  5. IMO Resolution MEPC.288(71). 2017 Guidelines for ballast water exchange (G6) imo.org
  6. IMO Resolution MSC.99(73). Amendments to SOLAS 1974, revised chapter II-2 imo.org
  7. MAIB Report 12a/2000. Capsize and sinking of the stern trawler Fraoch Ban gov.uk
  8. MAIB Report 13/2013. Capsize of the fishing vessel Sarah Jayne assets.publishing.service.gov.uk
  9. JAIC Final report on MV ESTONIA (December 1997), Part 3. Chapters 15 to 21, failure analysis, findings and conclusions shk.se
  10. Paris MoU Deficiency Codes. Overview of Deficiency Codes, 1 July 2024 maritime.public.lu