MARPOL Annex I Reg 26: Cargo Tank Size Limits

Regulation 26 of MARPOL Annex I is the clause that makes the rest of the deterministic outflow chain bite. Regulation 24 assumes a block of damage. Regulation 25 turns that block into a number, the hypothetical outflow $O_c$ for side damage or $O_s$ for bottom damage. Regulation 26 then forces the naval architect to chop the cargo block into tanks small enough that the number stays under a ceiling. Without Regulation 26, the outflow formulas would be an interesting calculation with no consequence. With it, they become a hard subdivision constraint that decides how many bulkheads a tanker carries and where they sit.

The rule binds oil tankers of 150 gross tonnage and above that fall inside the delivery brackets of MARPOL Annex I: broadly, tankers delivered after 31 December 1979, plus the older 1977-and-earlier ships caught by the contract and keel-laying conditions in paragraph 1.2. It does not apply to oil tankers delivered on or after 1 January 2010 (regulation 1.28.8). Those modern ships answer instead to the double-hull spacing in Regulation 19 and the probabilistic accidental oil outflow performance of Regulation 23. So Regulation 26 is, in 2026, a legacy-fleet rule, but it remains live regulatory text in the consolidated Annex and it shapes the design logic that the double-hull era inherited.

This article works through the ceiling, the volume caps, the tank-length formulas, and the worked numbers, then maps how the old deterministic regime and the modern double-hull regime divide the same problem. The numbers and formulas here are quoted from the consolidated MARPOL Annex I text and from the MARPOL Convention as amended; treat the consolidated edition carried aboard the ship and the flag Administration’s interpretation as controlling for any actual approval.

Why the 1973/78 drafters capped tank size

The rule answers a single question the 1973 conference set itself: when one tank opens to the sea, how much oil can it release before the spill becomes catastrophic. Before MARPOL, a large crude carrier could carry its cargo in a handful of vast tanks, and a single side breach could drain one of them entirely. The Torrey Canyon grounding in 1967 had released roughly 119,000 tonnes of crude from a 120,000-tonne cargo, and the Amoco Cadiz would lose its whole 223,000-tonne cargo off Brittany in 1978. Those losses were so near-total because the cargo block was coarsely subdivided. One opening, one tank, most of the oil.

The drafters’ move was to bound the worst single-compartment spill rather than to chase a zero-spill ideal. They wrote a damage box, assumed it could land anywhere, and required that the oil released from the tanks it opened stay under a fixed ceiling. The ceiling is an absolute volume, not a percentage of cargo, which is the choice that does the real work: a 250,000-tonne crude carrier and a 40,000-tonne product tanker face the same $30{,}000\ \text{m}^3$ ceiling, so the large ship has to subdivide its cargo far more finely in relative terms. A handful of giant tanks can’t pass; many smaller ones can.

The naval-architecture consequence is direct and visible on any pre-2010 tanker’s general arrangement. More tanks, each smaller, separated by more transverse bulkheads, with the cargo block split lengthwise by one or two longitudinal bulkheads into wing and center tanks. The classic three-across, multi-set arrangement that dominated tanker design from the late 1970s to the double-hull mandate is the geometric solution to Regulation 26’s arithmetic. Each added bulkhead is structural steel that earns no freight, so the design settles at the fewest tanks that still clear the ceiling, which is exactly the constrained-optimization problem the rule was built to impose.

Where Regulation 26 sits in the deterministic chain

The deterministic outflow regime is three regulations read as one. Regulation 24 is the damage assumption: it fixes a parallelepiped, a rectangular box, of assumed hull breach in three dimensions for side damage and two conditions for bottom damage. Regulation 25 is the outflow calculation: it slides that box along the full length of the ship, sums the cargo volume the box would release at each position, and reports the worst case as $O_c$ or $O_s$. Regulation 26 is the constraint: it states the ceiling those worst-case numbers must respect, and it gives volume and length caps on individual tanks that, taken together, are the practical means of staying under the ceiling.

The reason the rule reaches back into the damage box matters for understanding why the tank-length formulas look the way they do. Regulation 24’s side-damage box has a longitudinal extent $l_c$ of $\tfrac{1}{3}L^{2/3}$ or 14.5 m, whichever is less, and a transverse extent $t_c$ of $B/5$ or 11.5 m, whichever is less, measured inboard from the side at the summer freeboard level. The bottom-damage box runs $L/10$ longitudinally for the forward 0.3L and $L/10$ or 5 m elsewhere, with a transverse extent up to $B/6$ and a vertical extent $v_s$ up to $B/15$ or 6 m. Those dimensions are the yardstick. A tank longer than the damage box can be fully opened by a single hit. Keep tanks short relative to the box and a single collision opens fewer of them, so the released volume drops. Regulation 26’s length caps are the formal expression of that geometric idea.

The naval architect doesn’t get to design tanks first and check outflow later as a courtesy. The outflow calculation under Regulation 25 is run anywhere in the length of the ship, meaning every longitudinal position of the assumed damage box is tested, and the governing case is the maximum. That maximum is what Regulation 26 paragraph 2 caps. So in practice the architect iterates: lay out tanks, run the outflow sweep, find the position that produces the largest $O_c$ or $O_s$, and if it breaks the ceiling, add a bulkhead or shrink a tank and re-run. The length caps in paragraph 4 are a shortcut that keeps most arrangements compliant without exhaustive iteration, but the outflow ceiling in paragraph 2 is the binding test.

For the formal definitions of the damage box and the outflow sums, see Regulation 24 damage assumptions and Regulation 25 hypothetical outflow, which sit alongside Regulation 26 in the same chapter of Annex I. The point to carry forward is that $O_c$, $O_s$, $W_i$, $C_i$, $K_i$, $Z_i$, $b_i$, $l_c$, $t_c$, and $v_s$ are all defined upstream, and Regulation 26 reuses them.

The dependency runs one direction only, which matters for how an approval is built. Regulation 26 can’t be checked without first fixing the Regulation 24 box dimensions for the specific $L$ and $B$, then running the Regulation 25 sums against a candidate tank layout. Change the layout and the Regulation 25 numbers change, but the Regulation 24 box does not, because the box scales only with hull length and breadth, not with the internal subdivision. So the architect treats the damage box as a fixed stencil and slides it across successive tank layouts until the Regulation 25 outflow drops under the Regulation 26 ceiling. The wing-tank width $b_i$ is the one variable that appears in all three regulations at once: it sets the $K_i$ reduction in the Regulation 25 side-outflow sum, it sets the length allowance in Regulation 26 paragraph 4, and it’s measured against the Regulation 24 transverse extent $t_c$. Get $b_i$ right and the other constraints tend to fall into line, which is why practitioners size the wing width first.

The outflow ceiling: 30,000 m3 or 400 times cube-root DW

Paragraph 2 of Regulation 26 is the single binding sentence. Cargo tanks shall be of such size and arrangement that the hypothetical outflow $O_c$ or $O_s$, calculated under Regulation 25 anywhere in the length of the ship, does not exceed:

where $DW$ is the deadweight of the ship in tonnes at the summer load line. Read the formula in three pieces. There’s a floor of $30{,}000\ \text{m}^3$ that applies to every tanker in scope no matter how small. There’s a deadweight-scaled term $400\sqrt[3]{DW}$ that lets larger ships carry a proportionally larger permissible outflow, because a very large crude carrier physically can’t subdivide its cargo finely enough to hold the same absolute outflow a handysize tanker manages. And there’s a hard cap of $40{,}000\ \text{m}^3$ that the deadweight term is never allowed to push past, regardless of how large the ship gets.

Work out where the terms cross. The deadweight term equals the floor when $400\sqrt[3]{DW} = 30{,}000$, that is $\sqrt[3]{DW} = 75$, so $DW = 421{,}875$ tonnes. The deadweight term hits the cap when $400\sqrt[3]{DW} = 40{,}000$, that is $\sqrt[3]{DW} = 100$, so $DW = 1{,}000{,}000$ tonnes. The arithmetic tells you something about the rule’s design intent: the floor of $30{,}000\ \text{m}^3$ governs every tanker up to about 421,875 tonnes deadweight, and since almost no oil tanker ever built reaches even half that figure, the floor is the operative limit for the entire real fleet. The cube-root term and the 40,000 cap are there to handle hypothetical ultra-large carriers and almost never bind in practice. The largest tankers ever built, the Batillus-class ULCCs of the late 1970s, ran around 550,000 tonnes deadweight, which still sits below the 421,875-tonne crossover, so even they were held to the 30,000 ceiling.

The cube-root relationship is itself worth a pause. Deadweight scales roughly with displacement, which scales with the cube of a linear hull dimension. Take the cube root of deadweight and you recover something proportional to a length scale. So $400\sqrt[3]{DW}$ is, dimensionally, the rule saying “permissible outflow may grow in proportion to the linear size of the ship,” which is the same scaling that governs how cargo block length and tank count grow. The drafters of the 1973 Convention chose a cube-root law rather than a linear-in-deadweight law precisely so the permissible outflow wouldn’t balloon with ship size; a linear law would have let a 300,000-tonne ship spill ten times what a 30,000-tonne ship could, which would have defeated the point.

One subtlety: $O_c$ and $O_s$ are computed separately, and each must independently respect the ceiling. The side-damage outflow $O_c = \sum W_i + \sum K_i C_i$ tends to govern for ships with wide center tanks and narrow wings, because a side hit opens wing tanks fully and reaches center tanks only when the wing is shallow. The bottom-damage outflow $O_s = \tfrac{1}{3}\left(\sum Z_i W_i + \sum Z_i C_i\right)$ carries the $\tfrac{1}{3}$ factor that recognizes a grounding rarely empties a tank the way a side breach does, so $O_s$ is usually the smaller of the two for a single-hull ship. The architect checks both and the larger one drives the design.

Volume caps on individual tanks

Paragraph 3 puts ceilings on individual tanks, not just on the swept outflow. These are the caps that translate the aggregate outflow limit into a per-tank discipline a yard can build to.

The volume of any one wing cargo oil tank must not exceed 75 per cent of the hypothetical outflow limit from paragraph 2. For the bulk of the fleet held to the $30{,}000\ \text{m}^3$ floor, that means a single wing tank can be no larger than $0.75 \times 30{,}000 = 22{,}500\ \text{m}^3$. The logic is direct: side damage opens a wing tank, and if a single wing tank held the full $30{,}000\ \text{m}^3$, a single side breach could in principle release the entire permissible outflow from one compartment, leaving no margin. Capping the wing tank at 75 per cent of the limit keeps a single side-damage event below the aggregate ceiling even before the $K_i$ reduction factor is applied.

The volume of any one center cargo oil tank must not exceed $50{,}000\ \text{m}^3$. Center tanks get a higher absolute cap than wing tanks because they’re shielded by the wing tanks against side damage; a side hit has to penetrate past the wing tank’s inboard bulkhead to reach a center tank, and the $K_i = 1 - b_i/t_c$ factor in the side-outflow formula reduces or zeroes the center-tank contribution whenever the wing tank is wider than the damage box. The center tank’s exposure is mainly to bottom damage, where the $\tfrac{1}{3}$ factor already softens the outflow, so a larger absolute volume is tolerated.

There’s a relief clause for segregated-ballast ships. In segregated ballast oil tankers as defined in Regulation 18, the permitted volume of a wing cargo oil tank situated between two segregated ballast tanks, each exceeding $l_c$ in length, may be increased to the full hypothetical-outflow limit, provided the width of the wing tanks exceeds $t_c$. Read that carefully: it lets a protected wing cargo tank go up to $30{,}000\ \text{m}^3$ rather than $22{,}500\ \text{m}^3$, but only when ballast tanks longer than the damage box flank it and the cargo wing itself is wider than the side-damage penetration $t_c$. The ballast tanks act as sacrificial buffers; a side hit lands in clean ballast water, not cargo, so the cargo wing’s larger volume never actually spills in the assumed damage case. This clause is the first place Regulation 26 reaches across to the segregated-ballast regime, and it rewards arrangements that put ballast where damage is most likely.

Cargo tank length limitations

Paragraph 4 is the geometric heart of Regulation 26 and the part most often quoted in design reviews. The length of each cargo tank shall not exceed 10 m or one of the following values, whichever is greater. The 10 m floor means short tankers never have to subdivide below 10 m regardless of the formula result; below that length the rule stops biting.

The governing formula depends on the longitudinal-bulkhead arrangement, because longitudinal bulkheads are what split a wide cargo block into wing and center tanks and so determine how much cargo a given damage box can reach. Three cases.

Where no longitudinal bulkhead is provided inside the cargo tanks, the length limit is:

Here $b_i$ is the minimum distance from the ship’s side to the outer longitudinal bulkhead of the tank in question, measured inboard at right angles to the centerline at the summer freeboard level, and $B$ is the ship’s breadth. With no longitudinal bulkhead the tank spans the full beam, so $b_i$ takes its maximum and the bracket trends toward the $0.2L$ cap. That $0.2L$ cap is the headline number practitioners remember: a full-beam tank with no longitudinal subdivision can be at most one-fifth of the ship’s length.

Where a centerline longitudinal bulkhead is provided inside the cargo tanks, the limit is:

The centerline bulkhead splits the block into port and starboard tanks, halving the volume any single side hit can reach, so the rule allows a longer tank for the same outflow. Note this case has no explicit $0.2L$ cap written into it, but the geometry keeps it bounded.

Where two or more longitudinal bulkheads are provided inside the cargo tanks, the rule separates wing tanks from center tanks:

• For wing cargo tanks: $L_{\text{tank}} \le 0.2L$.
• For center cargo tanks the limit turns on the ratio $b_i/B$. If $b_i/B \ge \tfrac{1}{5}$, the center tank may be up to $0.2L$. If $b_i/B < \tfrac{1}{5}$, the center tank reverts to the bracketed formulas: $\left(0.5\,\tfrac{b_i}{B} + 0.1\right)L$ where no centerline bulkhead is provided, or $\left(0.25\,\tfrac{b_i}{B} + 0.15\right)L$ where a centerline bulkhead is provided.

The $b_i/B \ge \tfrac{1}{5}$ test is the same one-fifth that recurs throughout the deterministic regime, and it’s not a coincidence: $t_c$, the side-damage transverse extent, is $B/5$ or 11.5 m whichever is less. When the wing tank’s inboard bulkhead sits at least $B/5$ from the side, the side-damage box can’t reach past it into the center tank, the protection is complete, and the center tank earns the full $0.2L$ allowance. When the wing is narrower than $B/5$, the damage box can reach the center tank, so the center tank’s length gets pulled back by the same formulas that govern an unprotected tank. The length rule and the outflow rule share one geometric criterion, which is why a tank arrangement that satisfies the length caps usually satisfies the outflow ceiling without a separate fight.

Paragraph 4 closes with two operational provisions that are easy to overlook but are part of the rule. Where a cargo transfer system interconnects two or more tanks, valves or closing devices must separate the tanks and must be closed at sea, so the subdivision can’t be defeated by leaving cross-connections open. And piping running through cargo tanks within $t_c$ of the side or $v_c$ of the bottom must carry valves at the point it opens into any tank, kept closed at sea when the tank holds cargo, so a breached pipe inside the damage zone can’t drain an otherwise-intact tank. Both clauses recognize that the assumed-intact tanks in the outflow calculation are only intact if the architecture stops a single breach from cascading through shared piping.

Longitudinal bulkheads, segregated ballast, and the double hull

Regulation 26 doesn’t mandate longitudinal bulkheads outright; it makes them the cheapest way to comply. The length and volume caps are easiest to satisfy by splitting a wide cargo block with one or two longitudinal bulkheads, which is why the classic single-hull tanker arrangement of two wing tanks and a center tank across the beam, repeated down the length, became standard. The bulkheads do double duty: they shorten the reach of a side damage box and they create the wing-versus-center distinction the volume caps depend on.

The interaction with segregated ballast tanks runs deeper than the paragraph 3 relief clause. Regulation 18 requires segregated ballast capacity arranged to give a defined draught and trim without recourse to cargo tanks for water ballast, and it specifies that the ballast tanks be located to provide protective coverage against side and bottom damage. The protective-location provisions of Regulation 18 and the outflow rules of Regulations 24 to 26 were drafted to reinforce each other: ballast tanks placed where damage is statistically likely absorb the breach, so the cargo tanks behind them never enter the worst-case outflow. A tanker that satisfies Regulation 18’s protective-location coverage will generally find its Regulation 26 outflow sweep easier to pass, because the ballast tanks zero out the $W_i$ terms wherever they sit. The two regulations are different lenses on one design move: put clean water between the sea and the oil.

The double hull of Regulation 19 is where the deterministic regime’s logic gets absorbed into a prescriptive geometry. Regulation 19 requires oil tankers of 5,000 tonnes deadweight and above delivered on or after 6 July 1996 to have the cargo tanks protected by double sides and a double bottom, with the wing and double-bottom spaces meeting minimum width $w$ and height $h$ defined by deadweight-scaled formulas. For a tanker of 5,000 DWT and above, the wing space width is $w = 0.5 + DW/20{,}000$ m or 2.0 m, whichever is the lesser, but not less than 1.0 m, and the double-bottom height is $h = B/15$ or 2.0 m, whichever is the lesser, but not less than 1.0 m. Set that double-bottom height beside the Regulation 24 bottom-damage vertical extent $v_s = B/15$ or 6 m whichever is less, and the design intent is visible: the double bottom is sized to keep the assumed bottom damage box inside the ballast space, so $h_i \ge v_s$ makes $Z_i = 0$ and the bottom-damage outflow $O_s$ collapses to zero for the protected tanks. The double hull is the deterministic outflow logic frozen into a guaranteed clearance, so the architect no longer has to run the sweep tank by tank.

That’s why Regulation 26 carries its own sunset in paragraph 4.7: it does not apply to oil tankers delivered on or after 1 January 2010. For those ships, the double-hull spacing of Regulation 19 plus the probabilistic accidental oil outflow performance of Regulation 23 replace the deterministic ceiling-and-length regime entirely. Regulation 23 doesn’t cap a single worst-case outflow; it computes a mean outflow parameter weighted across a probability distribution of damage cases, and requires that mean to stay under a deadweight-scaled limit, with side and bottom damage handled by separate probabilistic densities per the MEPC.122(52) explanatory notes. The shift from Regulation 26’s deterministic worst case to Regulation 23’s probabilistic mean mirrors the broader move in ship safety from deterministic to probabilistic damage stability. So the regimes divide cleanly by delivery date: a tanker delivered before 2010 lives under Regulations 24 to 26; one delivered in or after 2010 lives under Regulations 19 and 23. A mid-life tanker fleet in 2026 is mostly the latter, but the former still floats and still gets surveyed against its as-built Regulation 26 arrangement.

The damage-stability requirements of Regulation 28 sit alongside all of this. Regulation 26 controls how much oil escapes; Regulation 28 controls whether the ship stays upright and afloat after the same assumed damage opens its tanks to the sea. A tank arrangement that passes the outflow ceiling can still fail the damage stability criteria if the flooded tanks shift the center of buoyancy badly, so the two checks run in parallel on the same damage assumptions. The naval architect satisfies both or redesigns.

Worked example: a 96,000 DWT product tanker

Take a single-hull product tanker of 96,000 tonnes deadweight, length between perpendiculars $L = 230$ m, breadth $B = 42$ m, with cargo arranged in five sets of tanks down the length, each set comprising two wing tanks and one center tank split by two longitudinal bulkheads. This is the classic pre-double-hull arrangement. We test the arrangement against the Regulation 26 length cap and the outflow ceiling.

Start with the outflow ceiling from paragraph 2. The deadweight term is $400\sqrt[3]{96{,}000}$. The cube root of 96,000 is about 45.79, so $400 \times 45.79 = 18{,}316\ \text{m}^3$. Compare to the $30{,}000\ \text{m}^3$ floor: the floor is larger, so the permissible outflow is $O_{\max} = 30{,}000\ \text{m}^3$, well under the $40{,}000\ \text{m}^3$ cap. Every $O_c$ and $O_s$ this arrangement produces must stay at or below 30,000.

Now the per-tank volume caps. The wing-tank cap is $0.75 \times 30{,}000 = 22{,}500\ \text{m}^3$; the center-tank cap is $50{,}000\ \text{m}^3$. Suppose each wing tank holds $7{,}480\ \text{m}^3$ and each center tank holds $7{,}495\ \text{m}^3$. Both sit comfortably under their caps, so paragraph 3 is satisfied on volume.

Next the length cap from paragraph 4. The ship has two longitudinal bulkheads, so we use the two-or-more-bulkheads case. The wing tanks get $0.2L = 0.2 \times 230 = 46\ \text{m}$. For the center tanks, compute $b_i/B$, where $b_i$ is the distance from the side to the outer longitudinal bulkhead, that is the wing-tank width. Say the wing tanks are 9.0 m wide, so $b_i = 9.0$ m and $b_i/B = 9.0/42 = 0.214$. Since $0.214 \ge \tfrac{1}{5} = 0.2$, the center tank also earns the full $0.2L = 46\ \text{m}$. With five tank sets over 230 m of cargo length, each set spans 46 m, exactly the cap. The arrangement is at the length limit, which is a tight but legal design; a yard would typically pull tank length to about 44 m to leave margin for the outflow sweep.

Now the outflow sweep, the binding test. Slide the Regulation 24 side-damage box, longitudinal extent $l_c = \tfrac{1}{3}L^{2/3}$ or 14.5 m whichever is less, along the ship. $\tfrac{1}{3}(230)^{2/3}$: $230^{2/3} \approx 37.9$, so $\tfrac{1}{3} \times 37.9 = 12.6$ m, which is less than 14.5, so $l_c = 12.6$ m. A side hit of that length opens one wing tank fully. The side-damage transverse extent is $t_c = B/5 = 42/5 = 8.4$ m or 11.5 m whichever is less, so $t_c = 8.4$ m. The wing tank is 9.0 m wide, wider than $t_c$, so the inboard bulkhead survives, the center tank is not breached, and $K_i = 1 - b_i/t_c$ goes to zero because $b_i = 9.0 \ge t_c = 8.4$. The side-damage outflow at this position is then just the wing-tank volume, $O_c = 7{,}480\ \text{m}^3$. If the box straddles two adjacent wing tanks along the length, the worst case opens both, giving roughly $O_c \approx 14{,}960\ \text{m}^3$. Both are well under the $30{,}000\ \text{m}^3$ ceiling.

The bottom-damage outflow $O_s = \tfrac{1}{3}\left(\sum Z_i W_i + \sum Z_i C_i\right)$ carries the one-third factor and, for a single-hull ship with $h_i = 0$ (no double bottom), $Z_i = 1$. A bottom box opening two wing tanks and one center tank at a transverse station gives $O_s = \tfrac{1}{3}(7{,}480 + 7{,}480 + 7{,}495) = \tfrac{1}{3}(22{,}455) = 7{,}485\ \text{m}^3$. Again well under 30,000. So this arrangement passes: $O_c \approx 14{,}960\ \text{m}^3$ worst case, $O_s \approx 7{,}485\ \text{m}^3$, both under the $30{,}000\ \text{m}^3$ ceiling, with the wing and center tank lengths at the $46$ m cap and volumes under the per-tank caps. The design is compliant with margin on outflow and tight on length, which is the usual signature of a competently subdivided single-hull tanker.

Run the same hull as a double-hull design under Regulation 19 instead and the bottom-damage outflow vanishes: a double bottom of height $h = B/15 = 42/15 = 2.8$ m, capped at 2.0 m, gives $h = 2.0$ m, and where the assumed bottom damage box stays inside that 2.0 m clean-ballast space, $Z_i = 0$ and $O_s = 0$. That collapse is exactly why the modern fleet abandoned the deterministic outflow sweep: the double hull guarantees the protection that Regulation 26 had to verify case by case. The reader can reproduce every number here from the formulas above and the assumed tank volumes; nothing depends on a proprietary table.

Worked example: a compliant tank versus a non-compliant tank

The first example passed comfortably, which hides where the rule actually bites. This second example contrasts two versions of one tank on the same hull so the constraint shows its teeth. Keep the 96,000 DWT hull: $L = 230$ m, $B = 42$ m, $O_{\max} = 30{,}000\ \text{m}^3$, wing-tank cap $22{,}500\ \text{m}^3$, center-tank cap $50{,}000\ \text{m}^3$.

Version A is a single-bulkhead arrangement. The yard wants fewer tanks, so it fits one centerline longitudinal bulkhead and stretches the cargo block into four long sets instead of five. Each tank now spans 57.5 m to fill the 230 m cargo length in four sets. Test the length cap from paragraph 4, centerline-bulkhead case: $L_{\text{tank}} \le \left(0.25\,\tfrac{b_i}{B} + 0.15\right)L$. With a centerline bulkhead and no wing tanks, the tank reaches the full beam, so $b_i = B/2 = 21$ m and $b_i/B = 0.5$. That gives $\left(0.25 \times 0.5 + 0.15\right) \times 230 = (0.275) \times 230 = 63.25\ \text{m}$. The 57.5 m tank is under 63.25 m, so it clears the length cap. So far so legal.

Now the outflow. With only a centerline bulkhead and no protective wing, a side hit lands directly in cargo. The side-damage box opens one half-beam tank of $57.5 \times 21$ m plan running the full cargo depth. Say that tank holds $13{,}200\ \text{m}^3$. The side-outflow factor $K_i = 1 - b_i/t_c$ doesn’t help here, because there’s no inboard wing bulkhead between the sea and this cargo, so the full tank contributes. If the damage box straddles the transverse bulkhead between two adjacent half-beam tanks, the worst-case side outflow is $O_c \approx 13{,}200 + 13{,}200 = 26{,}400\ \text{m}^3$, still under $30{,}000$ but with almost no margin. Push the tank to $58$ m and $14{,}500\ \text{m}^3$ each to carry more cargo and the straddle case gives $O_c \approx 29{,}000\ \text{m}^3$, brushing the ceiling. A single design tweak that looks harmless on the length cap walks the outflow right up to the wall. This is the version a flag surveyor flags: legal on paper, no reserve against the sweep finding a worse longitudinal position.

Version B is the protected arrangement. Add a second longitudinal bulkhead to make narrow wing tanks 9.0 m wide flanking a center tank, the same two-bulkhead geometry as the first worked example, and shorten the tanks back to the five-set, 46 m layout. Now $b_i = 9.0$ m, $b_i/B = 0.214 \ge 0.2$, so both wing and center tanks earn the full $0.2L = 46$ m. The wing tank is wider than $t_c = 8.4$ m, so a side hit stays in the wing: $K_i = 1 - b_i/t_c \to 0$, the center tank is shielded, and the worst-case two-wing straddle gives $O_c \approx 7{,}480 + 7{,}480 = 14{,}960\ \text{m}^3$. Same hull, same cargo length, half the worst-case outflow. The difference between Version A at $26{,}400$ to $29{,}000\ \text{m}^3$ and Version B at $14{,}960\ \text{m}^3$ is one longitudinal bulkhead and a handful of transverse ones. That gap is the rule paying for itself: the extra steel buys roughly $12{,}000\ \text{m}^3$ of outflow margin, which is the headroom an Administration’s surveyor wants to see before signing the International Oil Pollution Prevention Certificate.

Take Version A one step further into non-compliance to see the ceiling actually break. Fit two long center tanks of $58$ m and $16{,}000\ \text{m}^3$ each with the centerline bulkhead removed, so each tank now spans the full beam. The length cap, no-bulkhead case, is $\left(0.5\,\tfrac{b_i}{B} + 0.1\right)L$ capped at $0.2L = 46$ m. A 58 m tank already breaks that cap outright, before any outflow sum. And on outflow, a full-beam tank of $16{,}000\ \text{m}^3$ exceeds the wing-tank volume cap of $22{,}500\ \text{m}^3$ only narrowly per tank, but the side-damage straddle of two adjacent full-beam tanks gives $O_c \approx 32{,}000\ \text{m}^3$, over the $30{,}000\ \text{m}^3$ ceiling. The arrangement fails three ways at once: the 46 m length cap, the no-protection outflow ceiling, and any reasonable margin. The fix is mechanical, add bulkheads, which is the design pressure Regulation 26 exists to apply.

How the tank arrangement shapes the rest of the design

The Regulation 26 arrangement isn’t a standalone pollution rule; it sets the framework the rest of the tanker design hangs on. Tank count and bulkhead positions fixed by the outflow ceiling determine the cargo-handling plan, the number of pumps and lines, and the granularity of segregation a parcel tanker can offer. A five-set, three-tanks-across arrangement gives fifteen cargo tanks and so up to fifteen segregations before slop and ballast, which is why product tankers built to this logic could carry multiple grades.

The same bulkheads feed the loading and stability calculation. Each tank is a free-surface source, and the marine stability booklet and loading computer must account for every tank the Regulation 26 arrangement creates. More tanks mean more partial-fill conditions to check but also finer control over trim and heel, because the loader can shift small parcels between many compartments. The tank arrangement that satisfies outflow also defines the lightweight versus deadweight split, since the steel of every additional bulkhead is lightweight that doesn’t earn freight.

Survey and certification close the loop. The arrangement approved under Regulation 26 is recorded on the International Oil Pollution Prevention Certificate, the IOPP form B that lists, against checkboxes, whether the ship complies with Regulation 26 and Regulation 26.4. A port state control officer boarding a pre-2010 tanker checks the as-built tank arrangement against the IOPP supplement, and a deviation, a bulkhead removed in conversion, a cross-connection added without valves, is a deficiency. The tank plan is not just a design artifact; it’s a certified configuration the ship must maintain, and the wing/center distinction the volume caps create is carried forward into the discharge-control and segregation rules elsewhere in Annex I.

The double-hull-era equivalent records differently. A tanker delivered after 2010 carries on its IOPP supplement the Regulation 19 double-hull compliance and the Regulation 23 accidental-oil-outflow performance figures rather than a Regulation 26 outflow sweep. So a PSC officer reads a different page depending on the ship’s delivery date, but the question is the same: does the as-built cargo block hold the oil where the rule assumed it would.

Limitations

Regulation 26 is a deterministic rule built on a single assumed damage box, and that’s its main limitation. The Regulation 24 box is a fixed parallelepiped with dimensions scaled only by $L$ and $B$; it doesn’t represent the statistical spread of real collision and grounding penetrations, which is why the post-2010 Regulation 23 regime replaced it with a probability-weighted mean outflow. Any conclusion from a Regulation 26 sweep is a worst-case-of-one-assumption result, not an expected-outflow result, and the two can rank designs differently.

The rule applies only to the delivery brackets in paragraph 1 and not to oil tankers delivered on or after 1 January 2010. Applying Regulation 26 formulas to a modern double-hull tanker is a category error; that ship is governed by Regulations 19 and 23, and its outflow performance is assessed by a different method with different numbers. Conversely, citing Regulation 23 logic at a pre-2010 single-hull ship misreads its certificate.

The figures in this article are quoted from the consolidated MARPOL Annex I text. National implementing legislation, flag Administration interpretations, and recognized-organization rules can add detail, and some Administrations apply equivalent or stricter standards under the Regulation 3 to 5 exemptions and equivalents framework. The worked example uses assumed tank volumes and dimensions chosen to illustrate the method; it is not a model of any specific ship, and an actual approval runs the full outflow sweep across every damage position with the as-built tank geometry, not the three sample positions shown here. For any design or survey decision, work from the consolidated edition carried aboard, the approved stability and tank-arrangement documentation, and the flag Administration’s guidance, not from this summary.

See also

• MARPOL Annex I: the full oil-pollution-prevention annex Regulation 26 sits within.
• MARPOL Convention: the parent instrument and its amendment history.
• Regulation 25 hypothetical oil outflow: the deterministic $O_c$ and $O_s$ calculation whose result this rule caps.
• Regulation 24 damage assumptions: the parallelepiped damage box that the outflow sweep slides along the hull.
• Regulation 23 accidental oil outflow (probabilistic regime): the post-2010 probabilistic regime that supersedes the deterministic outflow ceiling.
• Regulation 19 double hull: the double-hull spacing that absorbs the Regulation 26 logic for modern ships.
• Regulation 18 segregated ballast tanks: the protective-ballast regime that interacts with the wing-tank volume relief.
• Regulation 28 damage stability: the parallel check on stability after the same assumed damage.
• Damage stability and probabilistic damage stability: the broader stability framework.
• Port state control: how the certified tank arrangement is verified in service.