Case study

Ballast Free-Surface GM Loss Case Study

Naval engineering case study on ballast free-surface effect, slack tanks, corrected metacentric height, heel response, roll-period change, ballast sequencing, and validation evidence.

A vessel can lose a large part of its initial stability without taking on more weight. The mechanism is free-surface effect: liquid in partly filled tanks shifts as the vessel heels, creating a heeling moment and reducing effective metacentric height. The problem is operational as well as hydrostatic. A loading plan may be acceptable when tanks are pressed full or empty, then become unsafe when several tanks are left slack during transfer.

This case study follows an offshore support vessel preparing for departure after ballast transfer. The loading computer initially shows acceptable stability, but the tank status in the calculation does not match the actual soundings. Three ballast tanks are slack. Correcting for free surface reduces the effective GM enough to require an operational hold.

The case is a screening calculation for engineering judgement. Final vessel decisions must follow the approved stability booklet, loading software, class rules, flag requirements, and company operating procedures.

Case Context

The vessel has completed cargo loading and is trimming for departure. Ballast has been shifted to improve propeller immersion and forward draft. A watch officer notes that the loading computer still shows the port and starboard wing tanks as pressed, but manual soundings indicate they are about half full.

ItemValue
Displacement mass\Delta=8000\ \text{t}
Uncorrected metacentric heightGM=0.85\ \text{m}
Minimum operating GM used for departure screen0.35\ \text{m}
Beam used for roll-radius estimateB=20\ \text{m}
Roll radius of gyration estimatek=0.38B=7.6\ \text{m}
Operational transverse heeling momentM_H=320\ \text{t m}
Departure heel-screen limit8^\circ
Actual tank statethree ballast tanks slack

The transverse heeling moment represents a combined small-angle screen for wind, minor cargo offset, and operational asymmetry. It is not a substitute for a full righting-arm curve or weather criterion.

Field Evidence

The issue is discovered before departure because the loading records do not agree:

EvidenceEngineering interpretation
remote level trend shows two wing tanks between 45 percent and 55 percenttanks are slack, not pressed
manual sounding confirms the remote level trendthe level signal is credible
ballast valve lineup shows a transfer was stopped mid-sequenceprocedural interruption created multiple slack tanks
loading computer status still marks tanks as pressedmodel state is wrong
vessel roll feels slow during berth movementlow effective GM is plausible
no structural overload or flooding alarm is presentproblem is stability margin, not hull damage

The most important finding is the mismatch between actual tank condition and the stability model. A stability calculation with the wrong tank state can be worse than no calculation because it creates false confidence.

Free-Surface Correction

The loading computer gives uncorrected:

GM=0.85\ \text{m}

The actual slack tanks have free-surface corrections:

Slack tankFree-surface correction
port wing ballast tankFSC_1=0.31\ \text{m}
starboard wing ballast tankFSC_2=0.24\ \text{m}
forepeak ballast tankFSC_3=0.15\ \text{m}

Total correction:

\displaystyle \sum FSC_i=0.31+0.24+0.15=0.70\ \text{m}

Corrected metacentric height:

\displaystyle GM_{corr}=GM-\sum FSC_i
GM_{corr}=0.85-0.70=0.15\ \text{m}

The corrected value is positive, but it is far below the departure screening minimum:

0.15<0.35\ \text{m}

This is enough to stop departure. A positive GM is not the same as an acceptable loading condition.

Heeling Response Screen

For a small-angle screen, use:

M_R\approx \Delta GM \tan\phi

Solving for heel angle:

\displaystyle \tan\phi\approx\frac{M_H}{\Delta GM}

With the corrected GM:

\displaystyle \tan\phi=\frac{320}{8000(0.15)}
\tan\phi=0.267

Therefore:

\phi=\tan^{-1}(0.267)=14.9^\circ

The estimated heel exceeds the departure screen:

14.9^\circ>8^\circ

If the original uncorrected GM had been used, the same heeling moment would give:

\displaystyle \tan\phi=\frac{320}{8000(0.85)}=0.0471
\phi=2.7^\circ

That difference explains why the tank-state error matters. The vessel did not become heavy; it became tender because the liquid was free to shift.

Roll-Period Screen

A simplified roll-period estimate is:

\displaystyle T_\phi\approx2\pi\sqrt{\frac{k^2}{gGM}}

Using:

k=7.6\ \text{m}

and corrected:

GM_{corr}=0.15\ \text{m}

gives:

\displaystyle T_{\phi,corr}=2\pi\sqrt{\frac{7.6^2}{9.81(0.15)}}
T_{\phi,corr}=39.4\ \text{s}

This is only a screening estimate, but it is consistent with a tender vessel. A very long roll period can be a warning sign when it appears after a ballast transfer and is not part of the planned loading condition.

The same estimate with uncorrected GM=0.85\ \text{m} gives:

\displaystyle T_{\phi,uncorr}=2\pi\sqrt{\frac{7.6^2}{9.81(0.85)}}=16.5\ \text{s}

Roll period alone is not a stability approval method, but the change reinforces the free-surface diagnosis.

Engineering Decision

The vessel should not depart in the actual tank condition. The engineering decision is:

Hold departure, restrict nonessential deck operations, correct the loading computer tank states, press or empty the slack ballast tanks according to the approved sequence, and release the vessel only after soundings, valve lineup, loading software, and stability criteria agree.

Immediate controls:

  1. stop ballast transfer until the officer, chief engineer, and master agree on the current tank state;
  2. enter actual tank soundings into the loading computer;
  3. avoid additional transverse cargo or crane operations while GM_{corr} is low;
  4. press up or empty tanks in a sequence that avoids creating more slack tanks;
  5. maintain port-starboard symmetry unless the stability plan explicitly approves otherwise;
  6. record final soundings and valve positions before removing the hold.

The operational issue is not solved by “adding more ballast” in a generic sense. Ballast placement and tank fill state decide whether the correction improves or worsens stability.

Corrective Ballast Sequence

The approved correction is to press up the two wing tanks and empty the forepeak to the required departure state. After completion, only a small service tank remains slack.

Corrected free-surface state:

Tank condition after correctionFree-surface correction
port wing ballast tank pressed0.00\ \text{m}
starboard wing ballast tank pressed0.00\ \text{m}
forepeak ballast tank emptied0.00\ \text{m}
service tank slack0.12\ \text{m}

New total correction:

\displaystyle \sum FSC_{new}=0.12\ \text{m}

New corrected metacentric height:

GM_{new}=0.85-0.12=0.73\ \text{m}

Heeling response under the same operational moment:

\displaystyle \tan\phi_{new}=\frac{320}{8000(0.73)}=0.0548
\phi_{new}=3.1^\circ

Roll-period screen:

\displaystyle T_{\phi,new}=2\pi\sqrt{\frac{7.6^2}{9.81(0.73)}}=17.8\ \text{s}

The corrected condition restores margin for the departure screen. The loading computer must still confirm all applicable criteria, including trim, draft, freeboard, downflooding, longitudinal strength, and any route or weather limits.

Validation Evidence

Release should require evidence, not only an instruction that tanks were corrected.

EvidenceAcceptance purpose
manual soundings for affected tanksconfirms actual liquid levels
remote level trendconfirms sensors agree with soundings
valve lineup recordconfirms transfer path is closed and stable
loading-computer printout or controlled electronic recordconfirms corrected tank state and stability criteria
total free-surface correction registermakes slack-tank contribution visible
departure draft and trim readingsconfirms the loading model matches the vessel
master/chief engineer sign-offassigns operational responsibility for release

If any of these disagree, the vessel remains in a restricted state. A single loading-computer green status is not enough when the inputs are uncertain.

Uncertainty Check

Assume the uncorrected GM has uncertainty:

u_{GM}=0.05\ \text{m}

and the combined free-surface correction has uncertainty:

u_{FSC}=0.08\ \text{m}

Combined uncertainty in corrected GM is:

u_{corr}=\sqrt{u_{GM}^2+u_{FSC}^2}
u_{corr}=\sqrt{0.05^2+0.08^2}=0.094\ \text{m}

The estimated corrected GM before action is:

GM_{corr}=0.15\ \text{m}

The lower bound is:

0.15-0.094=0.056\ \text{m}

The condition is not close to the departure requirement in a useful way. It is below the screening minimum even before considering uncertainty, and the uncertainty shows that the real condition could be extremely tender. The correct decision is operational hold and correction, not debate over a few centimetres of GM.

Risk Screen

A risk-priority-number screen helps document the operational hold:

RPN=S\times O\times D

Before correction:

RPN_{before}=9(4)(5)=180

Severity is high because low effective stability can lead to excessive heel, downflooding exposure, cargo movement, or loss of operational control. Occurrence is moderate because interrupted ballast transfers are credible. Detection is weak because the loading computer can look acceptable when tank states are wrong.

After corrected ballasting, tank-state verification, and release controls:

RPN_{after}=9(2)(2)=36

The severity of a future recurrence remains high, but the occurrence and detection scores improve when slack-tank combinations are controlled and recorded.

Lessons for Marine Operations

The transferable lessons are:

  1. Free-surface effect is an operational condition, not only a design note.
  2. Several modest slack tanks can remove most of the usable GM margin.
  3. Loading software is only as reliable as the tank states entered into it.
  4. Roll-period observations can support diagnosis but cannot replace stability criteria.
  5. Release after ballast correction requires soundings, valve lineup, loading records, and sign-off.

The engineering decision is to treat tank state as a stability-critical configuration. A vessel should depart only when the actual ballast condition and the approved stability model describe the same vessel.

REF

See also