Wind-Assisted Propulsion on Existing Ships: Retrofit Constraints, Savings Drivers and How EEXI Counts Wind

Does wind-assisted propulsion work on an existing ship?

It can, but the answer depends on your route, your deck layout and how the saving is measured — not on the technology alone. Lloyd’s Register describes wind-assisted propulsion technologies as able to “potentially offer double-digit fuel savings,” while DNV reports that vessel owners and operators have recorded fuel savings of between 4.5% and 9% in service, with potential savings of up to 25% if a system is installed as a retrofit, and margins said to be higher still on newbuilds designed around a sail system. Anemoi, the rotor-sail supplier on LR’s bulk-carrier projects, says that since installing the solution clients have seen up to 15% fuel savings. None of these numbers are interchangeable: they come from different parties — operators, a class society, a supplier — under different conditions, and the rest of this guide is about the factors that separate them on your ship.

The main WAPS families and what each needs from the ship

FuelEU Maritime (Regulation (EU) 2023/1805, Article 3(9)) defines wind-assisted propulsion as propulsion, whether partial or full, of a ship by wind energy harnessed by means of systems such as, inter alia, rotor sails, kites, hard or rigid sails, soft sails, suction wings or turbines. The IMO’s guidance on innovative technologies describes its own scope more narrowly: it applies to wind propulsion technologies that directly transfer mechanical propulsion forces to the ship’s structure. DNV describes the landscape as rigid or soft wing sails, Flettner rotors and ventilated foils, or kites, built on “a combination of advanced aerodynamics, automation, computer modelling and modern materials.” Lloyd’s Register states that wind-assisted propulsion does not require specialist crew competencies or additional crew numbers; DNV puts it in terms of the control system, noting that most modern systems use intelligent control and automation to operate safely without the requirement for additional crew. Treat both as statements about the technology in general, not as a manning conclusion for your ship — the chosen system’s operating procedures and familiarisation training, and any condition your flag or class attaches to the installation, are worth settling with the supplier, class and flag before the business case is signed off.

What every family needs from the hosting ship is structural and electrical capacity, not just deck area — but how much of each, and at which interfaces, is technology- and ship-specific. The integration package described below is what a rotor-sail retrofit on a bulk carrier required, and the scope moves with the configuration even within one technology: Lloyd’s Register notes that a hydraulic folding deployment system can also bring modifications to the ship’s deck hydraulic system. Read the list as an example of scope, not as your scope — that comes from a ship-specific assessment with the supplier, class and yard. In Lloyd’s Register’s bulk-carrier retrofit programme with Anemoi, the structural work centred on deck foundations and seats for the sails, plus cross-rail systems, and for structural continuity these foundation and seat structures should ideally be aligned with the ship’s primary structural elements, such as underdeck web frames and transverse bulkheads; detailed calculations and structural analysis were carried out to ensure the sail loads are efficiently supported by them. Deck outfitting items such as vent pipes and access walkways were relocated to make room. On the electrical side, modifications were needed to the ship’s electrical system including the main switchboard so the sail could draw power, along with deck wiring to run cable to the sail deck seat; the documentation package included an electrical system update, an electrical load balance update and short-circuit calculations. Cabling, connection boxes, the bridge control unit and an anemometer were all part of the installation. The specific deployment mechanism — sails folded from vertical to horizontal by hydraulic cylinders, or moved on rails by electric trolley or winch — also has to be chosen with the ship’s cargo-gear clearance and air draft in mind, since the potential for sail interference with shore-side and shipboard cargo gear, along with air draft limitations for port approaches, is a stated constraint.

What the saving depends on

The IMO’s own methodology for crediting wind-assisted propulsion (MEPC.1/Circ.896) ties the available power gain to three inputs: the ship’s reference speed, a wind propulsion system force matrix that expresses the force characteristic of the system against ship speed, wind speed and wind angle relative to heading, and a global wind probability matrix built from a statistical survey of worldwide wind data on the main global shipping routes. Note what that third input is and is not: it is one global statistical dataset, not your trade. The figure the method produces is therefore a fleet-generic wind credit, not a route-specific saving forecast — which is why a trade-specific estimate has to be asked for separately. The power the system itself consumes to operate is then subtracted from the gained wind power before any net benefit is counted — so a system with a high auxiliary draw gives back part of what it earns.

The IMO method is also deliberately simplified, and the simplification runs one way. MEPC.1/Circ.896 states that secondary effects of applying a WAPS which might increase ship resistance are ignored for the purpose of its guidelines — additional drag due to leeway, rudder angle and heel, and reduced propeller efficiency in light running condition — on the stated grounds that this is done without significant loss of accuracy. The corresponding forces are still considered, but for a different purpose: to rule out conditions that do not allow safe operation of the ship, for instance due to exceeding heel angles. So the credited figure and the in-service figure are answering different questions, and a crediting method is not a performance warranty. DNV, for its part, says uncertainties in operational cost remain with the weather, and that the advantages of sailing can be enhanced with weather routing algorithms that generate optimal routes for individual vessels. Taken together: ship speed, the WAPS’s own force characteristics, its own power demand and real wind exposure on the trade are what a saving estimate rests on — a single average percentage from a brochure is not a substitute for working through them for your ship.

Retrofit constraints to settle before the economics

Lloyd’s Register’s experience across three bulk-carrier rotor-sail retrofits with supplier Anemoi spans a wide size range: four prototype rotor sails of 2.1 m diameter and 16 m height on the geared Ultramax Afros in 2018, moved on longitudinal rails by electric trolley; three second-generation sails of 5 m diameter and 24 m height on the Kamsarmax TR Lady in 2023, deployed by electric winch on transverse rails in the cross-deck strips; and five sails of 5 m diameter and 35 m height, hydraulically folded, on a 400,000 DWT, 360 m Valemax very large ore carrier in October 2024 — which Lloyd’s Register’s August 2025 account described as the largest vessel to date to have WAPS installed, a ranking worth re-checking against a current source before it is quoted. Anemoi’s engineering design manager notes that each vessel differed significantly in the size, the deployment system and the number and size of the rotor sails. LR’s own conclusion is that close cooperation between class, technology provider and shipowner is what makes these conversions run smoothly, regardless of which wind technology is installed or what type of vessel it goes on.

The first lesson is sequencing. On TR Lady, the work was split into a “wind-ready” modification in November 2022 — all steel work and outfitting for the foundations and cross rails, plus cabling, connection boxes, bridge control unit and anemometer — and sail installation in June 2023. Because the bulk of the structural work was already complete, the lifting process took 1.5 days, spread over two calendar days: one rotor installed on the first day and the remaining two on the following day. Before that lift, a full-scale test site with a configuration identical to the shipboard installation was established, where the rotors were run at full speed and balanced and the lift and drive system tested; this land-based testing formed the basis of the Factory Acceptance Test (FAT), together with emergency stop, local control, bridge control and software settings. Phase-one preparatory work on the vessel included shimming the foundations and a survey confirming that levels and geometrical tolerances from the wind-ready phase matched the drawings. Once the sails were bolted down at the anti-uplift device locations and the umbilical cables hooked up to the deck boxes, a Harbour Acceptance Test (HAT) followed protocols similar to the FAT with the addition of vessel data, concluding once all checks were done on the rail deployment system, the interfaces between the ship’s data system and the sail control system, and the rotor sails while spinning.

The second lesson is that compliance should be checked only once an appropriately sized and well-operable configuration has been defined — because the regulatory impacts are specific to the type, size, deck position and number of rigs chosen. In all three cases, the regulations in force at the time of implementation, especially those on navigational safety, did not consider wind assistance rigs, so alternative solutions or exemptions had to be considered and justified. The examples are instructive because each fix was different. On TR Lady, rotor sails on the centreline obscured the bridge’s view of the forward blue Panama Canal light, which the Panama Canal Authority requires to be clearly visible along the centreline; the solution was port and starboard blue lights, as ACP vessel requirements allow. In two of the projects the sails would have obscured forward visibility of the aft main masthead light: on the VLOC a vertical extension of the main mast raised the light, while on TR Lady, where air draft and sail height were restricted by a Mississippi River bridge, a port and starboard split main masthead light was proposed and accepted by Marshall Islands flag.

The third lesson is the one most likely to be missed in a schedule: stability. LR calls the re-examination of ship stability arguably one of the more onerous compliance issues for a WAPS installation — lightweight and longitudinal and vertical centre of gravity (LCG and VCG) change, which engages the requirements of SOLAS II-1 on subdivision and damage stability. The VCG changed on both TR Lady and the VLOC, and on TR Lady lightship weight increased as well, so an inclining experiment had to be carried out. That is hard to do on a ship in retrofit and out of service: inclining experiments usually need calm conditions with no wind and are typically run overnight in dead-ship condition with loose moorings. On TR Lady the experiment was carried out at the wind-ready phase, in November 2022 — which is the point LR draws from it, that an inclining experiment can be scheduled into the wind-ready phase rather than left to the end. All modifications at the sail-to-ship interface are subject to approval and survey by class.

The practical takeaway for a first retrofit: settle the rig configuration and deployment system against your own cargo-gear and air-draft limits first, build the “wind-ready” integration work into a yard period you already have planned, and put the navigational-safety and stability sign-off in front of class and flag early, not as a late check before delivery. Whether an inclining experiment belongs in that package is a ship-specific question, not an automatic step — it follows from the weight and VCG change the installation causes and from what class and flag require by way of stability re-examination. Ask that question early enough that, if the answer is yes, the experiment can sit in the wind-ready phase as it did on TR Lady.

How the rulebooks count wind: EEDI/EEXI and FuelEU

Start by confirming that the index applies to your ship at all. Under MARPOL Annex VI as amended by MEPC.328(76), two regulations work as a pair: Regulation 23 requires the attained EEXI to be calculated for ships falling into the ship-type categories it lists, and Regulation 25 sets the required EEXI that the attained value must meet, for the same list of categories and for ships to which the chapter is applicable. What a surveyor checks is the pair — the verification that the ship’s attained EEXI is in accordance with the requirements in Regulations 23 and 25 takes place at the first annual, intermediate or renewal survey, or the initial survey, whichever is the first, on or after 1 January 2023, and after a major conversion of a ship to which Regulation 23 applies, a general or partial survey “shall ensure that the attained EEXI is recalculated as necessary and meets the requirement of regulation 25” — that qualifier is the treaty’s own wording, not a hedge, and the practical reading is that the surveyor decides the extent of recalculation while compliance with Regulation 25 is not optional. Regulation 19.3 then provides that Regulations 22, 23, 24 and 25 do not apply to ships which have non-conventional propulsion — defined in Regulation 2.2.19 as a method of propulsion other than conventional propulsion, including diesel-electric, turbine and hybrid propulsion systems — except that Regulations 23 and 25 do apply to cruise passenger ships having non-conventional propulsion and to LNG carriers with either propulsion type. Regulations 22 to 25 and 28 also do not apply to Polar Code category A ships. The 400 gross tonnage threshold in Regulation 19.1 is where the chapter starts, but it is not the whole applicability test: Regulation 19.2 takes the chapter’s provisions off ships solely engaged in voyages within waters subject to the sovereignty or jurisdiction of their own flag State — though each Party should still ensure such ships are constructed and operated consistently with chapter 4 so far as is reasonable and practicable — and off ships not propelled by mechanical means, platforms including FPSOs and FSUs, and drilling rigs regardless of propulsion. So the sequence is GT threshold, then trading pattern, then ship type and propulsion arrangement, confirmed with your flag State. The circular is candid about the same boundary: MEPC.1/Circ.896 states in its scope that it “might not provide sufficient measures of calculation and verification for ships with diesel-electric propulsion, turbine propulsion and hybrid propulsion systems” because the attained EEDI formula may not be able to apply to such systems. On a diesel-electric or hybrid ship, then, the circular is not a self-sufficient route to a wind credit — the applicability question goes to class, flag and verifier before any calculation starts.

For EEDI and EEXI, the IMO’s MEPC.1/Circ.896 (14 December 2021, superseding MEPC.1/Circ.815) sorts innovative energy-efficiency technologies into categories A, B and C. Wind assistance — the circular’s own examples are sails, Flettner rotors and kites — sits in Category B-2: a technology that reduces propulsion power at V_ref but does not generate electricity, and that can be used at its full output only under limited conditions, so its availability factor (f_eff) is less than 1.00, unlike a Category B-1 technology for which f_eff is treated as 1.00. Read that as the category description it is, not as a dial: for a WAPS, f_eff is not chosen independently — see the matrix method below.

Note how much narrower that box is than the FuelEU definition quoted earlier. The circular’s WAPS section applies to wind propulsion technologies that directly transfer mechanical propulsion forces to the ship’s structure (sails, wings, kites), and those are the technologies its B-2 list names. The distinction is not whether the device loads the ship — a turbine’s aerodynamic loads reach the hull through its mounting like any other rig, which is why its foundation, lateral loads and VCG effect need the same structural and stability treatment. The distinction is what the wind energy is converted into. A turbine — which FuelEU’s definition does include — does not deliver mechanical propulsion force to the structure; it generates electricity, and Lloyd’s Register notes that a shipboard wind turbine is more commonly used to generate auxiliary power than propulsion, providing electricity for onboard systems and so reducing the load on the diesel generators, adding that space, stability and efficiency concerns mean turbines are not typically used as the primary means of propulsion. The circular’s own table puts emission reduction achieved by generating electricity for the normal maximum sea load on the auxiliary-power side, as Category C, counted as P_AEeff. But that only tells you which side of the table to look at: the circular’s Category C annex carries calculation and verification methods for two technologies only — waste heat recovery (C-1) and photovoltaic power generation (C-2). There is no wind-turbine method in it. So a turbine or any wind-driven generating set must not be dropped into a B-2 calculation by analogy, and it does not acquire a ready-made Category C method either. The category, the calculation method and the verification basis all have to be fixed with the verifier first, or the credit will not survive review.

The quantity actually credited is the available effective power, written as f_eff × P_eff — the two are combined because the product of availability and power is the result of a matrix operation. It is built from three pieces: the ship’s reference speed (V_ref) in knots; a wind propulsion system force matrix F(V_ref)k for the specific system and ship, each element giving the propulsion force in kN for the respective wind speed and angle, the angle in relative bearings with 0° on the bow; and the global wind probability matrix W_k, whose elements give the probability of each wind speed and relative angle and sum to 1. “For the specific system and ship” is the circular’s default, not a stylistic flourish: it requires the benefit of any innovative technology to be evaluated in conjunction with the hull form and propulsion system it is intended for, and warns that model-test or sea-trial results obtained with a different hull form or propulsion system may not be applicable. The one relaxation it offers is the multi-unit case described below. A fourth matrix, P(V_ref)k, carries the system’s own power demand in kW and is subtracted from the gained wind power. The formula also applies η_D, the total efficiency of the main drive(s) at 75% of the rated installed power (MCR) of the main engine(s), which “shall be set to 0.7, if no other value is specified and verified by the verifier”.

Two mechanics inside that formula are worth knowing before you read a supplier’s number. First, the summation in the formula does not run over the whole matrix. Each element of W_k is the probability of one wind speed and relative angle, and the sum over all elements of W_k equals 1 — the full set of wind conditions. But the formula for (f_eff · P_eff) carries two side conditions: “sorting all force matrix elements in descending order”, and q — “the number of elements added in the formula” — is defined so that the probabilities of the first q−1 elements sum to less than one half while the first q elements sum to one half or more. The sum is then normalised by the probability actually included. So the credited figure is drawn from the highest-force elements that together make up the first half of the wind-probability mass, not from the whole matrix. Second, the matrix is bounded by the system’s real operating envelope. F(V_ref)k must be zero for any pair of wind direction and wind force not in conformity with the operational domain of the system validated by the verifier in the operations manual of the system and the ship. Two further conditions sit in the circular’s own steps for calculating the force matrix terms: the potential wind drag induced by the system is to be accounted for, such as in unusable wind directions close to head wind and when the system is not operational due to exceedance of operational limits; and if F(V_ref)k exceeds the resistance of the ship, such that the propeller thrust would be negative, F(V_ref)k is to be limited at the resistance value.

The force matrix itself is established by one of three routes — a wind tunnel model test, a CFD/numerical calculation, or a full-scale test — and the choice is not free: for some system types wind tunnel tests are not appropriate, in which case numerical calculation is accepted provided it is referred to experimental results and carried out to ITTC 7.5-03-01-02 and 7.5-03-01-04 or equivalent, and if neither is suitable another method needs the verifier’s approval. Whichever route is used, the procedure must be submitted to the verifier in advance of conducting the test or calculation, with the detailed report following afterwards. Where multiple units are fitted, forces may be determined for the devices in isolation and summed only if a validated method accounts for interaction between the propulsors and between ship and propulsors.

The structural and stability side is inside the index work too, not bolted on afterwards. The lateral forces and yawing moments the system applies to the ship are to be documented by the builder or manufacturer and observed by the verifier, and conformity with the operational domain requires that, for any wind direction and force pair, the strength of the system, the forces at the embedment and the list of the ship conform with the ship’s structural design file and stability file. Where lateral forces and yawing moment are particularly significant, the verifier may request course-keeping and rudder-angle demonstrations. At final verification, the total net power generated by the system is confirmed on the basis of the documentation in the EEDI Technical File, and the verifier checks that the configuration of the installed system agrees with the system described in that file. The practical conclusion is about sequence: settle the rig type, size, number and deck position before the Technical File is built on them, and agree the handling of any later change with your verifier and class before the change is made, not after.

FuelEU Maritime obligations only start if the ship is in scope, and that scope is drawn quite differently from the EEXI scope above — higher tonnage threshold, no ship-type list, but tied to EU ports. Under Article 2(1), Regulation (EU) 2023/1805 applies to ships above 5,000 gross tonnage that transport passengers or cargo for commercial purposes, regardless of flag — to the energy used during a stay in an EU port of call, all of the energy on voyages between EU ports of call, and half of the energy on voyages to or from a non-EU port of call (and on outermost-region legs). Article 2(7) takes warships, naval auxiliaries, fish-catching and fish-processing ships, wooden ships of a primitive build, ships not propelled by mechanical means, and government ships used only for non-commercial purposes out altogether, and Article 2(3) to (6) let Member States exempt certain island, outermost-region and public-service passenger routes until the end of 2029. A ship outside all of that has no FuelEU monitoring-plan duty at all.

For a ship that is in scope, the regulation works with two quantities of its own, P_Wind and P_Prop, defined in its Annex I. The operative requirement for a technical manager is Article 8(3)(p): for a ship equipped with wind-assisted propulsion, the monitoring plan must contain a description of the installed wind propulsion equipment on board and the values of P_Wind and P_Prop. A retrofit triggers this directly — Article 9(2)(b) requires the plan to be modified without undue delay when a new energy conversion system enters into use, Article 9(3) requires proposed modifications to be notified to the verifier without undue delay, and Article 11(2) makes that class of modification subject to verifier assessment, after which the verifier notifies the company whether it conforms. Once satisfactorily assessed, the plan is recorded in the FuelEU database and is accessible to the administering State (Article 11(3)) — note that this is an assessment of conformity by the verifier, not an approval by the administering authority. The regulation’s stated approach is to recognise sources such as wind in the ship’s compliance balance through approximations of their contribution, precisely because measuring them accurately is difficult — and point (p) is where a wind installation shows up in the monitoring plan.

That approximation is where business cases go wrong, so be precise about what the benefit is. The regulation’s own reasoning is that wind “greatly reduces the GHG intensity of the overall ship energy use” and that the difficulty of measuring it should not impede its recognition “through means of approximations of their contribution to the ship’s compliance balance” — in other words, what the ship earns is an effect on its calculated GHG intensity, reached through the Annex I methodology from the declared equipment and its P_Wind and P_Prop values, not from the aerodynamic output of the rig as such. So before a figure goes into a business case, work it through Annex I in the current consolidated text with your verifier: those same P_Wind and P_Prop values are what the verifier will assess when it reviews the monitoring plan.

Adjacent topics, each covered separately: FuelEU Maritime compliance strategy, EU ETS shipping compliance guide, IMO Net-Zero Framework guide, FuelEU penalty estimator, CII rating estimator, and for where wind assistance sits among other retrofit options, the existing-fleet maritime tech frontier. One point from the circular does cross over: if more than one innovative energy-efficiency technology is subject to approval in the same certification, interactions between them have to be considered, and the appropriate technical papers should be included in the additional information submitted to the verifier in the certification process.

A screening checklist for the first conversation with a supplier

• Route data: Can the supplier model your saving against your actual trade’s wind exposure, rather than only the generic global wind probability matrix used for the EEXI calculation?
• Wind-force calculation method: Will the force matrix come from a wind tunnel model test, a CFD/numerical calculation, or a full-scale test — and is the supplier prepared to submit that test or calculation procedure to your class verifier in advance of carrying it out, as MEPC.1/Circ.896 requires?
• Operational domain: What are the system’s operational limits, including the maximum operational wind speed, and are they written into the operations manual the verifier will validate? Outside that envelope the method credits zero force, and the drag of a shut-down rig still counts against you.
• Lateral loads and stability: Has the supplier documented the lateral forces and yawing moments, and have they been checked against your ship’s structural design file and stability file? Ask what the installation does to lightweight, LCG and VCG — on both TR Lady and the VLOC the VCG changed, and on TR Lady lightship weight increased too — and ask early whether your flag and class will require an inclining experiment, and when it can be run.
• Class approval route: Who is managing structural, electrical and navigational-safety approval with your class and flag — the supplier, the yard, or your office?
• Deck and bridge sightlines: Does the proposed rig’s height, position and deployment system risk obscuring a required navigation light — masthead, or a canal transit light — or interfering with cargo gear and air draft, and if so, who designs the alternative arrangement and secures flag acceptance?
• Electrical integration: Besides the sail’s own supply, who produces the electrical load balance update and short-circuit calculations for the modified switchboard?
• Auxiliary power draw: What is the system’s own power demand — since that figure is subtracted from the gained wind power before any saving is credited under the IMO method?
• Configuration lock: Is the rig type, size, number and deck position settled? Final EEDI/EEXI verification checks the installed configuration against the Technical File, and FuelEU wants the installed equipment and its P_Wind and P_Prop values in the monitoring plan.
• Index applicability: Does MARPOL Annex VI Regulation 23/25 apply to this ship at all, given its gross tonnage, trading pattern, type and propulsion arrangement — and does FuelEU’s Article 2 scope catch it? If the ship has diesel-electric, turbine or hybrid propulsion, settle with class and flag how the wind credit is to be calculated and verified before relying on MEPC.1/Circ.896 alone.
• Performance verification (contractual, not statutory): How will delivered performance be verified after installation — for example, on-off testing, the subject of a published Lloyd’s Register whitepaper on performance verification of wind-assisted ship propulsion systems — and will the supplier document it? This is a commercial and operational check on what you bought; it is separate from the EEDI/EEXI final verification, which is a documentation and configuration check.

Decision path: screen → study → class/flag → install → verify

Step	What happens	Who owns it
1. Screen	Match candidate WAPS types against route, deck layout, cargo-gear and air-draft limits; confirm which index obligations the ship actually falls under, given its gross tonnage, trading pattern, type and propulsion arrangement	Your office
2. Study	Force matrix established by wind tunnel, CFD or full-scale test; available effective power (f_eff × P_eff) estimated; lateral forces and yawing moments documented	Supplier, with the verifier receiving the test or calculation procedure in advance
3. Class/flag approval	Structural, electrical, stability and navigational-safety modifications reviewed and approved; alternative light arrangements justified	Class and flag administration
4. Install	”Wind-ready” structural/electrical integration — any inclining experiment class and flag require can sit here — then sail installation, preceded by land-based FAT and followed by HAT	Supplier and yard
5. Verify (statutory)	Net power confirmed from the EEDI Technical File documentation and installed configuration checked against that file; and, if the ship is in FuelEU scope, the monitoring plan modified with the installed equipment and its P_Wind and P_Prop values and notified to the verifier for assessment	Class/verifier and your office
5b. Verify (performance — separate)	Delivered performance measured against what was sold, e.g. by on-off testing. Not part of the EEDI/EEXI final verification under MEPC.1/Circ.896 — a contractual or operational exercise	Your office, with supplier and class

Each step produces the input the next one needs. The table is a screening order, not a published programme or a universal project plan. The real order of work will move with the class society, the flag administration, the yard and the technology; what LR draws from the three projects is narrower and more general, that close cooperation between class, technology provider and shipowner is what makes these conversions run smoothly, regardless of which wind technology is installed or what type of vessel it goes on. The circular does enforce part of the sequence, though: the test procedure goes to the verifier before the test, and final verification compares what is on deck against what was approved.