8.1. The Importance of Tank Size

The single most important structural factor in reducing pollution in the event of damage is the one that's almost never talked about: tank size and arrangement. IMO has developed a method to evaluate a tanker's propensity to spill cargo in the event of hull damage. It is based on a hypothetical collision and a hypothetical grounding. The collision involves a wedge penetrating into the hull about B/5 where B is the ship's beam. The grounding involves damage from the bottom up to about B/15. These parameters were based on a study of past casualties. The collision/grounding is assumed to be equally likely to occur anywhere along the ship's side/bottom. The overall result of this analysis is the ship's Effective Oil Spill (EOS) number, which is the percentage of the ship's cargo which will be spilled on average given the IMO collision/grounding scenario. The IMO system is far from perfect but it is a reasonable starting point for evaluating a design's resistance to spillage.

Table 8.1 shows the IMO Effective Oil Spill numbers for four different pre-MARPOL VLCC's and ULCC's all built in the mid 70's, a typical single hull MARPOL VLCC built in 1986, a 2001 built double hull VLCC, and the Hellepont 440,000 ton double hullULCC. From age 25 on, the pre-MARPOL ships must operate under either IMO REG 13G4 (30% of the side or bottom tanks non-cargo) or IMO REG 13G7, usually known as Hydrostatically Balanced Loading (HBL). For these ships, Table 8.1 shows the EOS numbers for each of these regimes.
DesignPercent Cargo SpilledNumber of
As-built13G713G4Cargo Tanks
Hellespont Embassy, 19762.11.82.2|34 five across
Empress des Mer, 19762.62.22.435 three across
Shell L-Class, 19743.63.13.722 three across
Ludwig VLCC, 19743.42.93.322 three across
Typical Marpol Single Hull, 19864.313 three across
SHI 1321, Double Hull VLCC, 20012.42.1*17 three across
DSME 5184 Double Hull ULCC, 20021.91.6*21 three across
* Bottom outflow reduced by 50% for oil captured in double sides.

Table 8.1 makes a number of points including:

  1. In terms of expected spillage, 13G7 is clearly superior to 13G4. 13G4 involves keeping a number of tanks totally empty, which lifts the ship out of the water, while the remaining tanks are filled right up to the brim. From a spill minimization point of view, this is exactly the wrong way to go. (13G4 also involves higher loss in carrying capacity than 13G7, less flexibility in multi-parcel loads, very high stresses, and most importantly putting ballast in tanks that were not designed to handle sea water. Corrosion in unprotected tanks as a result of conversion to segregated ballast was at a minimum a strong contributory cause to the Erika sinking and spill. If enough ships go 13G4, we will see more Erikas. [After this was originally written, the Prestige spill occured with almost exactly the same causation as the Erika.]) Spill minimization requires keeping the cargo as low in the water as possible in order to reduce the outflow head. As Table 8.1 shows, often a ship will have a higher EOS number under 13G4 than if IMO Regulation 13G never existed.

    Despite this some large charterers, notably BP, adopted a policy of only chartering 13G4 ships, putting strong pressure on the owners to adopt 13G4 rather than 13G7. These companies thought they would get more flack from the media if they had a spill on an HBL ship than on a segregated ballast ship. A cynical decision that did not work out for Total.
  2. The MARPOL tankers have terrible EOS numbers. Like 13G4 ships, MARPOL tankers operate with about a third of their tanks completely empty and the rest filled to the brim. The difference is that the designers reacted to the MARPOL requirements by making the ships very tall and the tanks very big, both of which exacerbate spillage further. One can argue that this was an acceptable price to pay to obtain segregated ballast. We find such an argument unconvincing. The nearly new Exxon Valdez would have spilled a lot less oil if she had been an older ship.
  3. The EOS number for a new double hull VLCC fully loaded is about 2.4%. For all the problems with the double hull, we end up with a ship that spills more oil percentage-wise in the IMO damage scenario than a good pre-Marpol ULCC.

But for present purposes, the interesting feature of Table 8.1 is the wide range in expected spillage under 13G7 for the pre-Marpol ships. The Hellespont Embassy spills 65% less than the L-class in the same casualty scenario. The other two pre-Marpol ships are in between with the Empress des Mer much closer to the Embassy and the Ludwig V's much closer to the L-class.

This is a product of small tank size. The Embassy and the Empress have a lot more tanks than the other two designs. The Empress, which was built to the last pre-Marpol restriction on tank size, has 13 pairs of wing tanks; the Shell L-class has 8. And in the case of the five across Embassy, the tanks are much more intelligently arranged.

An interesting feature of double hulls is that they finesse the IMO tank size limits. The IMO tank size limits apply only to cargo tanks that touch the side shell plate. Double hull wing tanks are thus exempted and the tanks in modern double hulls are enormous. The standard newbuild VLCC with a cargo cubic of 350,000 m3 has only 15 cargo tanks plus two small slop tanks. Even the L-class, which we've just finished castigating, with a cargo cubic of 385,000 m3 has 22 cargo tanks. The Marpol VLCC's were even worse. Most of these ships were built with 13 cargo tanks. If one were determined to increase spillage in a casualty via structural regulation, it's hard to imagine a more effective set of requirements than the Marpol single hull rules.

The main reason why the double hull ULCC is superior to the double hull VLCC is that is has 14 wing tanks while the standard double hull VLCC has 10 wing tanks. All it took was two extra wing tank bulkheads to reduce expected spillage by more than 30%.

8.2. Double hull versus single hull

Table 8.2 shows a little more detail on the Embassy operating under hydrostatic balance versus new double hull comparison.
2002 ULCC2001 VLCCHellespont Embassy (13G7)
SIDE LOSS(m3) 16,282 15,330 12,195
BOTT LOSS(m3) 5,336 3,384 6,783
The double hull cuts bottom loss because the probability of the bottom damage extending into a tank is reduced but the new ships get clobbered on side damage due to the massive wing tanks. Since the double sides are only about 3 meters wide, the reduction in probability of side damage penetrating into a cargo tank is overwhelmed by the increase in outflow if the tank is penetrated.

This is a little unfair to the double hull because in a grounding a portion of the bottom loss will be captured in the top of the double sides. IMO arbitrarily and without any support says cut the bottom loss number for double hull by 50%. Adopting this rule, the double hull V's EOS number becomes 2.1%, the asterisked number in Table 8.1. But even if we do that, the double hull VLCC still has a 17% higher Effective Oil Spill number than the Embassy under 13G7. Paradoxically, double sides are more effective in containing grounding damage than they are in reducing side damage. The way to go after side damage is sub-division. And the double bottom does almost nothing in grounding that could not be done far more efficiently by HBL.

8.3. Sloshing in Massive Tanks

Sadly Class is implicated in the growth of double hull tank size. The biggest tanks in a modern double hull VLCC are about 50 meters long. Tanks this long have a natural sloshing period which is close to the ship's natural pitch period, 13 to 15 seconds. When a tank's sloshing period matches the ship's pitch period, immense waves can build up in the tank, crashing from end to end. This is known as sloshing resonance.

In the mid-70's no one would be crazy enough to design a tank to operate anywhere near sloshing resonance. Thus, even if an owner cared nothing about pollution -- and few did -- you either built smaller tanks or you had to use real (complete) swash bulkheads. The difference in cost between a complete swash bulkhead and a oil-tight bulkhead is not all that great, so the incentive to go with massive tanks was not strong. But over the years, Class has allowed the swash bulkhead to atrophy and then disappear. The argument is that we can operate these tanks at resonance because we can predict the forces and beef up the structure to handle them.

This argument is a sad joke on a number of levels. Nobody has any way of accurately estimating sloshing forces at resonance, certainly not Class. The best of the current lot of Class tools is probably LRFLUIDS. When Daewoo applied LRFLUIDS to the case of the Hellespont double hull ULCC's center tank, the program indicated that at resonance the tie beams would impose an important dampening on the sloshing. ABS's empirical gouge said the same thing. The state of the art in sloshing analysis is the Hamburg Ship Research Institute's program which implements a full two phase Navier-Stokes but only 2-D. Despite being 2-D, this is an extremely computationally intensive program. One run simulating a little over one minute in real time took a cluster of 8 Dual-Pentium PC's over two days to compute. The results showed that at resonance the tie beams will have almost no effect on sloshing. The basic wave form is a kind of U that sneaks under the tie-beams as it moves from one end of the tank to the other, not a sort of semi-harmonic wave as Class claims. (Any housewife who has had water slosh out of a basin could have told us the same thing.) The Hamburg results are far closer to reality but the people at Hamburg will be the first to tell you that they cannot accurately predict the loads imposed on the structure. But when this wave crashes into a bulkhead, it climbs over 15 m into the top of the tank, dwarfing the wave heights in the 2005 tsunami.

The only reasonable thing to do is to stay away from resonance. And that means real swash bulkheads, not overgrown webs. And once you have real swash bulkheads, the move to full oil-tight bulkheads and far better sub-division is obvious. The result is a big reduction in spillage and far more flexibility in cargo parcels, ballast exchange, and tank inspection.

The tragedy is that this big improvement would cost very little. Shorter tanks mean:

  1. sloshing forces are markedly reduced which saves steel,
  2. much more importantly, a far more even distribution of transverse forces to the side shell and far better ability to withstand asymmetric loading and racking forces.
An egg carton is a very efficient structure. Cut out every other transverse corrugation and then see what you have to do to get the strength back. Because of the increased efficiency of the structure, limiting tank size doesn't require that much extra steel. The Empress' lightweight, 60,656 tons, is nearly the same as other U's of the same size with larger tanks built a year or two earlier. The Embassy's lightweight, 57,628 tons, is actually a bit smaller than other good pre-Marpol tankers of her size but it is easily the best tanker structure we have ever seen. At time she was scrapped at age 27, we had not found one crack in the Embassy. She is a very instructive exception to the even-good-ships-have-a-few-cracks rule enunciated in Section 4. In other words, small tank size is not expensive. Our kind of regulation: substantial effect on spillage at nil resource cost.

In the case of the massive double hull tanks, we have just the opposite effect: immense forces in the critical lower hopper area near the transverse bulkheads, and in way of the centerline buttress, and at the stringer corners and web toes, despite the Rules' staying away from any really difficult loading condition. The bottom bracket of the centerline buttress on the DSME ULCC has a 57 mm web and a 60 mm faceplate, both High Tensile Steel. The center tank lower web toes have 50 mm faceplates. That's way too much stress in one place.

8.4. Obvious Conclusion

At a minimum, the IMO Regulation 24 tank size limits should be applied to all double hull wing tanks and the IMO Regulation 13G(7) prohibition against operating tanks near sloshing resonance should be applied to all tankers.