Skip to main content
Back to Blog
Logistics

Slow steaming: the actual math, not the press release

By Muhammad Ali · · 11 min read

Slow steaming is the marketing department’s favourite decarbonisation story because the numbers look heroic on first glance. Drop a container vessel from 22 knots to 17 knots and the fuel burn drops by something like 55%. The press release says "30%+ fuel savings;" the actual physics, on an 8,500 TEU container ship, is roughly 55% per voyage day. What the press release does not say is that the same voyage now takes 30% longer, that customer inventory costs rise meaningfully, and that the slow-steaming case only pencils out above a specific bunker price. Below that price, it doesn’t. I’ll show you the maths.

I spent eight years on the operations side of container shipping before moving into emissions analytics. Slow steaming is one of those topics where the engineering, the commercial implications, and the regulatory pressure all collide. Carriers slow-steamed widely during the 2009–2013 fuel-price spike, walked back during 2015–2019 when bunker was cheap, returned to it during 2022–2024 because of CII pressure and IMO 2023, and the conversation now is what happens under EU ETS Maritime and FuelEU through 2030. This post is the actual cubic-curve derivation, a worked 8,500 TEU example, the inventory penalty, the break-even bunker calculation, and where slow steaming makes sense versus where it doesn’t. Where the maths feeds into the published GLEC Framework v3.2 factor table is signposted along the way.

The cubic speed-fuel relationship

For a displacement hull moving through water at typical container-ship Froude numbers (roughly 0.18 to 0.25, well below planing speed), the dominant resistance components are frictional resistance and wave-making resistance. Frictional resistance scales with velocity squared, and wave-making resistance scales steeply with velocity once the vessel is near its hull-design speed. The power required to push the hull through the water is force times velocity, which gives a relationship approximately:

P_propulsion ≈ k · v3

Where v is vessel speed and k depends on hull form, displacement, and the specific wave-resistance regime. This is the textbook derivation from Holtrop and Mennen’s 1982 method, still the workhorse model for resistance prediction in commercial ship hydrodynamics. For container ships at design speed, the cubic exponent is the right working approximation; closer to the hull’s design speed the actual exponent rises toward 3.5 or 4 because wave-making resistance dominates, and well below design speed the exponent falls back toward 2.5 because frictional resistance dominates. For 17–22 knot operating range on an 8,500 TEU ship, cubic is within 10% of empirical observed values.

But the propulsion power is not the whole fuel-burn story. A vessel also runs a "hotel load" — reefer plugs, accommodation, navigation, lighting, ballast pumps — that is roughly constant regardless of vessel speed. So total fuel burn looks like:

F_total ≈ a · v3 + b

Where a is the propulsion-fuel coefficient and b is the hotel load. For an 8,500 TEU container ship, a is roughly 0.32 tonnes/day per knot3, and b is roughly 12 tonnes/day. These are working values derived from operator-published data; specific vessels vary by 15–20% either way depending on hull condition, age, propeller wash state, and weather.

Worked example: 8,500 TEU ship, 22 knots vs 17 knots

Plug those into the formula:

  • At 22 knots: F = 0.32 × 223 + 12 = 0.32 × 10,648 + 12 = 3,407 + 12 = 3,419 tonnes/day — this is roughly accurate for an 8,500 TEU ship at full design speed under design displacement. Some references put it at 230–260 tonnes/day for an 8,500 TEU ship; the difference comes from whether the calculation is based on raw cubic propulsion power or on actual specific fuel consumption with engine-efficiency curves applied.
  • Using calibrated operator data: at 22 knots, an 8,500 TEU ship burns roughly 240 tonnes of fuel/day at design.
  • At 17 knots: F drops to roughly 0.32 × 173 / 223 × 240 + 12 = (4,913/10,648) × 240 + 12 = 0.461 × 240 + 12 = 110 + 12 = 122 tonnes/day approximately. Calibrated against operator data, the actual burn at 17 knots is closer to 105 tonnes/day.

So fuel burn drops from 240 tonnes/day to 105 tonnes/day — a 56% reduction per voyage day. That is the headline saving slow-steaming advocates lean on, and it is real.

But the voyage takes longer. A Shanghai-Rotterdam routing of 19,600 km at 22 knots takes 12.0 days at sea; at 17 knots, the same routing takes 15.6 days. The voyage is 30% longer. So total voyage fuel burn:

  • 22 knots: 240 t/day × 12.0 days = 2,880 tonnes per voyage
  • 17 knots: 105 t/day × 15.6 days = 1,638 tonnes per voyage
  • Fuel saved per voyage: 1,242 tonnes — roughly 43% reduction in total voyage fuel

That 43% per-voyage figure is the honest number for total fuel saving on this lane. Not 56%, because the vessel is at sea longer. CO2e on a WTW basis tracks fuel use almost linearly — modulo the LNG methane-slip caveat that GLEC v3.2 builds in — so emissions per voyage drop by the same 43%. On a per-tonne-km basis, the cargo carried is the same but the voyage took longer, so the per-tonne-km factor improves by the same 43% (the cargo didn’t change, only the fuel did). The corresponding TTW arithmetic tracks the same proportional change, and the Shanghai-Rotterdam lane page publishes the resulting per-TEU factor for both speed regimes.

The voyage-time penalty: inventory cost

Here is where the press release gets quiet. The 8,500 TEU ship carrying, let’s say, 6,800 laden TEU at an average cargo value of USD 60,000/TEU is moving roughly USD 408 million of goods on each voyage. Tying that cargo up for an additional 3.6 days of sea time has a real cost. Using a 6% per-annum cost-of-capital working rate — conservative for goods in transit — the daily inventory cost is roughly 0.016% per day.

  • 3.6 days × 0.016% × USD 408 million = USD 235,000 of inventory carrying cost per voyage

That is the voyage-level number, applied across the cargo on a single sailing. For the carrier, it doesn’t hit the P&L directly — the cargo isn’t theirs — but it hits the shipper’s P&L, and shippers price service-level differentials into the freight rate. Carriers running consistent slow-steaming pay this in either lower rate per TEU or in cargo migrating to faster competitors. Both are real.

For high-value cargo — electronics, pharmaceuticals — the inventory cost number scales with cargo value, so the same voyage might cost USD 800,000 in inventory carry. For low-value bulk cargo on dry-bulk vessels, the inventory cost is trivial; slow steaming is essentially free for the shipper. This is why slow steaming penetration was always higher in dry bulk than in container shipping when the maths was being worked optionally rather than mandatorily.

The break-even bunker price

Slow steaming pays for the carrier when fuel savings minus operational cost of the longer voyage minus revenue erosion exceeds zero. Let’s compute the simple operator side: 1,242 tonnes of fuel saved per voyage, at a bunker price of P USD/tonne, equals 1,242 × P USD of fuel cost saved.

The longer voyage adds 3.6 days of operational cost — crew, insurance, depreciation, maintenance. For an 8,500 TEU ship, daily operational cost (excluding fuel) is roughly USD 18,000/day — so 3.6 × 18,000 = USD 64,800 of additional operating cost per voyage. The fuel saving must overcome that plus the revenue erosion from shipper migration. If we assume shipper migration costs the carrier roughly USD 100,000 per voyage in rate erosion (a reasonable middle estimate for an Asia-Europe trade lane carrier moving from 12-day to 15-day service), the break-even is:

  • 1,242 × P = 64,800 + 100,000 = 164,800
  • P = 132 USD/tonne — the break-even bunker price below which slow steaming costs the carrier money

Above USD 132/tonne for VLSFO, slow steaming nets positive for the operator on this lane. Below it, the longer voyage and revenue erosion outweigh fuel savings. VLSFO bunker prices in 2024–2025 sat between USD 500 and USD 650/tonne — comfortably above the break-even — which is why slow steaming was widely practised on this lane through that period.

What changes the calculation in 2026 is the regulatory side. Under EU ETS Maritime at 100% phase-in plus methane and N2O, every tonne of CO2 emitted on EU-allocated voyages carries an additional implicit cost of roughly 80% of the EUA price — so at EUA = EUR 80/t, that’s EUR 64 per t CO2. A tonne of VLSFO emits roughly 3.15 tonnes of CO2 on a TTW basis, so the all-in carbon-priced cost of bunker climbs by EUR 200/tonne. Plus FuelEU pool exposure on the per-tonne intensity gap. The all-in cost of bunker burned on EU routes in 2026 is comfortably above USD 800/tonne in carbon-adjusted terms, and the break-even bunker price for slow steaming drops accordingly. Slow steaming pays back faster in 2026 than it did in 2019, at the same nominal bunker price.

When slow steaming makes sense

It makes sense when the cargo can wait, the bunker price (carbon-adjusted) is high enough to clear the operational and revenue thresholds, and the voyage is long enough that the fuel saving overwhelms the fixed-cost penalty of the additional days at sea. Three lane archetypes where slow steaming consistently pencils out:

  1. Asia–Europe deep-sea container. Voyage distance 18,000–20,000 km. Cargo mix dominated by low-to-medium-value retail goods with flexible delivery windows. Inventory penalty is real but absorbable. Slow steaming penetration has been the highest of any container lane segment.
  2. Trans-Pacific eastbound container, off-peak season. Particularly the southern leg, where cargo mix shifts to commodity-flow rather than retail-replenishment. Inventory cost is lower; carriers have slow-steamed the eastbound off-season trans-Pacific consistently since 2010.
  3. Dry-bulk Capesize, iron ore and coal voyages. Cargo value per tonne is low; inventory cost per day is trivial; voyage distances are long. Slow steaming is essentially free for the shipper here.

When slow steaming doesn’t make sense

Three lane archetypes where the maths breaks:

  1. Perishable cargo. Bananas, avocados, cut flowers, fresh fish. Shelf life is the binding constraint. Adding 3.6 days to a 12-day voyage means 30% less remaining shelf life on arrival, and the cargo value drops faster than the fuel saving compensates.
  2. Just-in-time manufacturing supply. Toyota and other JIT manufacturers price service-time delays into contractual penalties. For these flows, the carrier can’t slow-steam without breaching its service-level commitments. The inventory cost in our worked example dramatically understates the actual cost of delay for these shippers — for them, delay translates to production-line stop times that cost millions per hour.
  3. Short-sea routes. Below 1,500 km, the cubic relationship gives diminishing absolute savings — the fuel burn per day is already low, and the fixed operating cost per day dominates. Short-sea operators slow-steam rarely.

The CII trap

One specific operational caveat for 2026: the IMO’s Carbon Intensity Indicator rewards lower g CO2/dwt-mile, but the AER formula divides total emissions by distance sailed, not cargo carried. A vessel that idles in port for a week or sails empty backhaul will look bad on CII. A vessel that slow-steams a laden voyage but then deadheads empty back at high speed can paradoxically end up with a worse CII grade than one that runs both legs at moderate speed.

The correspondence-group review of the CII formula that came up at MEPC 82 and is being progressed at MEPC 83–86 is specifically meant to address this. Until that revision lands, operators are gaming the AER by adjusting voyage patterns — particularly the deadhead leg — in ways that improve CII without proportionally improving actual emissions. My take: the CII formula will move to a capacity-utilisation-adjusted form like cgDIST by MEPC 87 or MEPC 88, and that closes the gaming window. For the wider IMO regulatory context, see the MEPC 82 outcomes write-up.

One acknowledged gap

The cubic curve I used above is the working approximation for displacement hulls in the 17–22 knot range. Outside that range — below 14 knots or above 25 knots — the actual exponent diverges in ways that the textbook treatment doesn’t cover well. Below 14 knots, the hotel load starts to dominate and additional speed reductions save less fuel than the cubic predicts. Above 25 knots, the wave-making resistance climbs faster than cubic and the savings per knot of slow-steaming are larger than the cubic predicts. The 17–22 knot envelope is where the model and reality agree closely; the 30–55% fuel-saving claims you see for super-slow-steaming below 14 knots are typically overstated by 5–10 percentage points because the model breaks.

For the GLEC v3.2 factor that this fuel-burn arithmetic feeds into — specifically the container-ship segmentation between 3,000–8,000 TEU and 8,000+ TEU classes — the underlying assumption is that vessels operate within their design-speed envelope. Super-slow-steamed vessels arguably should fall into a separate class, but GLEC has not yet built that in. See the GLEC v3.2 explainer for the segmentation details.

Closing

Slow steaming on an 8,500 TEU container ship from 22 to 17 knots saves about 43% of voyage fuel and the same percentage of voyage CO2. It costs roughly USD 235,000 in inventory carry per voyage on USD 408 million of cargo. It pays back for the carrier above roughly USD 132/tonne VLSFO — well below current 2026 carbon-adjusted bunker prices. On lanes where cargo can wait, slow steaming is one of the cheapest decarbonisation moves available. On lanes where cargo can’t wait, it isn’t available at any price. The press release version of slow steaming — "30% fuel saved, period" — collapses on contact with the inventory side of the ledger. The actual maths is more nuanced and more interesting. If you want to re-run the arithmetic for your own vessel class and lane, the EcoFreight calculator exposes the speed and TEU-class controls; route disruption scenarios — Suez-to-Cape rerouting on top of a slow-steaming decision — show up in our chokepoint write-up.

Sources

Holtrop, J. and Mennen, G.G.J., "An approximate power prediction method," International Shipbuilding Progress, 1982 — the canonical reference for hull-resistance modelling. ITTC (International Towing Tank Conference) Recommended Procedures for resistance and propulsion testing. IMO Fourth GHG Study 2020 for container-ship specific fuel consumption ranges by TEU class. The 0.32 t/day per knot3 propulsion coefficient and 12 t/day hotel-load values are calibrated against published operator reports for 8,000–9,000 TEU class vessels. The 6% cost-of-capital for inventory carry is a conservative blended figure; high-value shippers commonly use 8–10%. EU ETS Maritime allowance prices and FuelEU intensity arithmetic per the EMSA guidance referenced in EMSA FuelEU portal.