The Prize: How Much of Resistance Is Friction

At 12.5 knots, our reference ship, a 208,000 dwt Newcastlemax bulk carrier, pushes 1,120 kN of calm-water resistance. Six hundred and seventy-six of those kilonewtons are skin friction: sea water shearing along 22,030 m² of painted steel. That is 60 per cent of everything her engine works against, and the largest single prize anywhere on the ship.

Which is why air lubrication attracts the claims it does. A 12.9 metre flat plate in the United States Navy's Large Cavitation Channel, with air injected underneath it, has been measured at better than 80 per cent drag reduction over the whole of its length. A vendor brochure offers up to 10 per cent fuel. On most of the runs of the first full-scale trial to measure a net saving, on the Japanese training ship Seiun Maru in 2001, the ship went slower with the air on than with it off. All three are honest, and all three describe the same physics at different points along a chain of multipliers. Building that chain is the whole of this lesson, so that you never again meet one of those numbers without asking which point it came from.

Multiplier one: how much of the resistance is friction

Frictional resistance is the streamwise force produced by wall shear stress integrated over the wetted surface, and on a slow full-form hull it dominates. Our Newcastlemax is 292.0 m between perpendiculars, 50.0 m in the beam, 18.30 m draught, block coefficient 0.85, displacing 232,800 t. At 12.5 knots (6.430 m/s) her length Reynolds number is 1.58 × 10⁹, the ITTC-1957 line gives a friction coefficient CF of 0.001448, and her total resistance coefficient is about 0.00240. Of that total, 0.001448 is bare-hull skin friction (60 per cent), 0.000362 is the form or viscous pressure term k·CF with a form factor k of 0.25 (15 per cent), 0.00040 is the roughness and correlation allowance ΔCF that any full-scale prediction carries (17 per cent), and the remaining 0.00019 is wave-making (8 per cent).

Wave-making is negligible here because she is slow for her length. Her Froude number is

Fn = V / sqrt(g · L) = 6.430 / sqrt(9.81 × 292.0) = 0.120

where V is speed in metres per second, g is 9.81 m/s² and L is the length between perpendiculars in metres. Below about Fn 0.15 a hull's wave system is cheap; above about Fn 0.25 it is ruinous, because it grows far faster with speed than friction does. That single fact sets the market for this technology.

Indicative shares of calm-water resistance taken by skin friction with its roughness allowance run 75 to 85 per cent for a VLCC, Suezmax or Capesize, 70 to 80 per cent for a handysize or MR product tanker, and 45 to 60 per cent for a large container ship at 21 to 23 knots. Treat those bands as orientation, not data: they vary widely with hull form and loading condition, and only your own resistance prediction is fit to design to. Those bands include the roughness allowance. The chain below takes the smooth-hull skin friction alone as its first multiplier, 60 per cent on our ship against 77 with the allowance, and that is a conservative choice: the roughness part of the allowance is wall shear too, and a layer that holds removes it over the plating it covers, but how much of a tank's allowance is roughness and how much is correlation with its own past ships is not something this course can split.

The second thing to fix is what a quoted drag reduction actually means. Every paper in the field uses

%DR = (1 − τw / τw0) × 100

where τw is the local wall shear stress at one station with gas injected and τw0 is the shear at the same station and speed without it, both in pascals. That is a local quantity on a covered surface. It is not a reduction in total resistance, not a reduction in delivered power, and certainly not a reduction in fuel. Nearly every disappointment in this technology's history traces back to treating those four things as the same number.

Multiplier two: only the flat bottom can be lubricated

Air rises. That is the one piece of physics working in your favour, and it only works where the plating is horizontal and above the air. Our ship has roughly 11,200 m² of flat of bottom, of which a realistic release-unit layout covers about 9,000 m². Against 22,030 m² of wetted surface that is 41 per cent, generous for any hull. A 2026 study for the European Maritime Safety Agency (EMSA), carried out by DNV, puts the flat of bottom of tankers, bulk carriers and gas carriers at typically 30 to 40 per cent of the wetted surface, and of the largest container ships at 25 to 35 per cent, laden, before anything is taken off for what a layout of release units can actually reach. Our ship, with 51 per cent of flat bottom and 41 per cent covered, sits above that range, at the optimistic edge of what a hull of her form offers (lesson 3 says why). A Newcastlemax is still the textbook candidate for area, and, as lesson 5 shows, a hard one for the blower: the same study judges the systems generally less suited to deep-draught ships such as large bulk carriers and tankers.

Take, as a theoretical example, a finer hull whose layout can reach only about 18 per cent of her wetted surface. Identical technology, less than half the multiplier, purely because of hull shape.

The flat of bottom is the limit for a second reason. An air layer is a two-dimensional sheet with two open edges, and containing it sideways is as hard as generating it. A laboratory plate is fenced in by the walls of its tunnel or by side plates; a ship's flat bottom is open at both sides. The clearest full-scale lesson is the cement carrier Pacific Seagull, reported by ABS at about 1 per cent net on her first trial, because the bubbles did not cover enough of the bottom, and at 10 per cent in ballast after end plates were welded along each side of the bottom to keep them in. Deadrise, curvature and a small flat of bottom are therefore not minor penalties: they are disqualifying. Lateral containment at ship scale, not the physics of the layer, is the single largest unresolved uncertainty in this subject.

Multiplier three: the drag reduction is not one hundred per cent

Now assemble the chain:

ΔR / RT = (frictional share) × (covered fraction) × %DR / 100

For the reference ship the first two terms are 0.60 and 0.41, so 24.7 per cent of her total resistance is friction sitting on covered plating. That assumes the local shear on the flat bottom matches the hull average, a simplification worth naming out loud.

Then the third term. At the 85 per cent a clean laboratory plate achieves, the ceiling is 21.0 per cent of total resistance, or 235 kN. Nobody has measured that on a ship. At a service assumption of 40 per cent, the saving is 676 × 0.41 × 0.40 = 111 kN, that is 9.9 per cent of total resistance, worth 111 × 6.430 = 714 kW of effective power and, at a quasi-propulsive coefficient of 0.68, 1,050 kW of delivered power. At 30 per cent it is 83 kN and 786 kW; at 60 per cent, 166 kN and 1,572 kW. The 40 per cent is an assumption, not a measurement, and where it lands is the argument of the next four lessons.

Only then does the blower bill arrive. The ship needs roughly 1.1 m³/s of air at the plating, delivered against 1.84 bar of hydrostatic head at 18.30 m draught, which absorbs something near 506 kW. Lesson 5 does it properly. Subtract it from the 1,050 kW and 544 kW survives, 5.1 per cent of her 10,590 kW of delivered power, from a physical prize that started at 60 per cent of resistance.

Be careful what you believe about that last figure. There is a full-scale record, and it is larger than the sceptics allow: ABS's 2019 advisory Air Lubrication Technology counts eighteen full-scale performance tests between 2002 and 2015, with reported net savings running roughly 4 to 10 per cent. By July 2025 more than 300 ships in service carried a system, by the count in EMSA's study. What that record is not is well measured. The advisory notes that the trial data were mostly not collected or analysed to the standard an EEDI trial demands; EMSA's study still calls for long-term measurement in service before credible saving figures can be set; and, as far as the published reports show, almost none of those tests recorded where the air actually went under the hull. The individual results run from Seiun Maru, whose runs mostly lost speed because the bubbles added resistance and reached the propeller, to Pacific Seagull at 10 per cent in ballast once her bottom was fenced in; and what separates them is where the air went, not the physics of the layer. So the 10 per cent in a brochure is a vendor figure, a 10 to 20 per cent quoted at full scale is a reduction in frictional resistance rather than in fuel (ABS lists one, on an air-cavity inland vessel), and the honest range for a net saving on a hull nobody has instrumented is nearer 0 to 8 per cent. The width of that band is why the fourth course in this sequence exists.

Monday morning

Do the three-multiplier sum on your own ship before a vendor does it for you. Take the skin friction share from your resistance prediction (smooth-hull friction alone, as this lesson does, for the conservative figure, and say so if you count the roughness allowance as well), the flat of bottom from the docking plan and the wetted surface from the hull form, and multiply. If the covered fraction comes out below about 25 per cent, or if skin friction carries less than about half her resistance, the prize is too small to be worth the parasitic load and you can stop there. If it survives that test, the next question is what %DR you are entitled to assume, and that depends entirely on which of three flow regimes the system actually runs in, which is the next lesson.