Vessel Insights: External Methodology

1. Introduction

Fleet managers, chartering teams, and counterparties all depend on accurate, up-to-date vessel data, yet vessel particulars are often scattered across brokers, questionnaires, class societies, P&I clubs, and correspondence, with no single version anyone can fully rely on.

Vessel Insights (VI) is Veson Nautical's authoritative Single Source of Truth (SSOT) for vessel particulars, ownership and management information, and operational data. It consolidates and reconciles data drawn from multiple proprietary sources and enriches it with modelled outputs, so users get one accurate, trustworthy record per vessel rather than having to reconcile conflicting information themselves.

2. Data Sources

VI draws on several distinct types of data, each contributing different strengths, which are combined to build a comprehensive and commercially relevant vessel record.

• Researched and validated vessel intelligence data: vessel identity, lifecycle milestones, structural particulars, ownership, and cargo capacity, researched and maintained on a daily basis by Veson's vessel data specialists. Spans more than approximately 65,000 vessels globally across the dry, wet, and gas segments.

• Owner and operator-submitted questionnaire data: vessel particulars submitted directly by owners and operators in the course of commercial operations, making this data operationally validated and widely trusted by charterers and vetting professionals. Particularly authoritative for fields such as TPC (tonnes per centimetre immersion), cargo tank capacities, classification society, and net register tonnage (NRT).

• Chartering correspondence data: vessel particulars that circulate in everyday shipping correspondence, systematically extracted to give a real-world, operationally current view of vessel specifics. Daily extraction makes this source especially valuable for catching recent changes to a vessel's details.

• Modelled and derived data: for vessel particulars where empirical values aren't available from the sources above, modelled estimates fill the gap, including speed and consumption curves across fuel types and load conditions. These models were first developed in 2021 and are continuously refined using market, emissions, and vessel movement data.

3. How the Data is Compiled

• Matching: records from each source are matched to the same real-world vessel using its IMO number, then unified into a single canonical entry. Duplicate or conflicting identities are resolved at this stage.

• Format and type validation: every value is checked against its expected data type and format before use. Values that don't conform are excluded from the final record or held for review, rather than published.

• Business rule validation: a comprehensive set of maritime logic checks is applied, for example, ensuring winter DWT is less than summer DWT, and that launch dates precedes delivery dates. These rules are developed and maintained by Veson's maritime domain experts.

• Source prioritisation: where more than one source populates the same field, an established prioritisation framework selects the most reliable value, based on which source is authoritative for that particular field.

• Modelled data integration: modelled speed and consumption values are added to each vessel record, and cross-referenced against empirical data where it exists.

• Refresh cadence: the dataset is refreshed daily.

• Ongoing monitoring: automated checks run on every daily refresh to catch anomalies before they reach the published dataset, including unexpected changes, drops in field coverage, or values falling outside expected ranges, all of which are flagged for review before being released to clients.

4. Data Quality & Validation

Every daily refresh of VI passes through a structured set of automated checks:

• Completeness: confirms each field is populated wherever a value is expected. Fields that only apply to certain vessel types (for example, gas-specific fields) are correctly excluded from this check where their absence is expected.

• Data type and range validation: confirms each field holds a value of the correct type, and for numeric fields, that it falls within defined minimum and maximum bounds — catching outliers and data errors.

• Schema integrity: confirms field names and structures match the published field specification.

• Source attribution: for every value in the dataset, the contributing source is recorded, giving full lineage and traceability back to its origin.

• Business rule validation: the maritime logic checks described in Section 3 are re-applied as a further quality gate.

These automated checks are supplemented by periodic review from Veson's maritime data specialists, who bring domain expertise that automated checks alone cannot capture.

5. Coverage & Scope

VI covers the commercial maritime fleet with IMO numbers, across the following segments:

• Bulk carriers (Handysize to Capesize)

• Tankers (crude, product, chemical)

• Container vessels

• Gas carriers (LNG and LPG)

• RORO / ROPAX

• General cargo and multi-purpose vessels

Fleet coverage spans approximately 65,000 vessels globally with active commercial activity.

Vessels without IMO numbers (such as barges) and non-cargo-carrying vessels (such as offshore and work vessels) fall outside the current scope of the dataset.

6. Carbon Intensity Indicator (CII) Methodology

6.1 What CII Measures

The Carbon Intensity Indicator (CII) is the IMO measure of a vessel's operational carbon efficiency. The most widely used form of CII is the Annual Efficiency Ratio (AER): the ratio of CO₂ emitted by a vessel to the transport work it performs (cargo capacity × distance travelled) over a calendar year.

Since 1 March 2018, vessels of 5,000 gross tonnage or more have been required to report this data to the IMO's Data Collection Service. Since January 2023, these vessels have also received an annual letter rating from 'A' (most efficient) to 'E' (least efficient); a vessel rated 'D' for three consecutive years, or 'E' for one year, must submit a corrective action plan.

6.2 How VI Estimates CII

Where a vessel's actual reported emissions data isn't available to a client, VI provides an independent estimate of CII/AER, built using AIS positional data and vessel specifications, following the calculation approach set out in the IMO's 4th Greenhouse Gas Study. In outline:

• Distance travelled is derived from the sequence of a vessel's AIS position reports.

• Fuel consumed and CO₂ emitted are estimated from an established naval-architecture relationship between engine power, vessel speed, and draft, combined with a fuel-efficiency curve for the main engine and a separate, steady-rate estimate for auxiliary engines (generators, boilers).

• The resulting attained AER is converted to a letter rating using the boundaries and annual reduction factors set out in the IMO regulation.

VI provides per-vessel estimates of attained AER and rating estimated attained AER and rating, CO₂ emitted, fuel consumed, distance travelled, and time at sea, covering the last three calendar years plus a running year-to-date figure, updated each time the vessel completes a port call.

6.3 Data Quality for CII Estimates

Because AIS signals can contain errors, VI applies data cleansing before they're used in a CII estimate, for example, discarding any pair of position reports that would imply a vessel travelling faster than its rated speed. Reported draft, which is sometimes temporarily misreported (for example, immediately after leaving port), is smoothed to its most frequently reported value over each journey, to avoid distorting laden/ballast status. A vessel's activity when stopped (for example, engaged in cargo operations versus idle) is inferred from position, speed, and draft together.

More broadly, steps taken to manage estimation uncertainty include cross-checking data from multiple sources, identifying and removing outliers, applying known physical limits, and subject matter expert review of outputs.

As with any modelled estimate, some uncertainty is inherent. At the time of writing, no comprehensive independent, publicly available emissions dataset exists against which CII estimates can be fully benchmarked; Veson will compare model outputs against such data and report accuracy figures if and when it becomes available.

7. Consumption Methodology

7.1 What Consumption Data Represents

VI provides modelled speed-consumption curves for vessels, covering laden and ballast draft conditions and HFO/LFO fuel types (see the MDO note in Section 7.4 below). This is modelled data, derived from a physics-informed model built and maintained by Veson's Data Science team since 2021, rather than data observed directly from a specific vessel's operations.

7.2 How VI Models Consumption

Rather than treating fuel consumption as a black box to be learned purely from data, VI's model encodes the physical relationships already well understood from naval architecture, for example, that main engine power scales approximately with the cube of vessel speed, and broadly linearly with draft, and uses AIS data to estimate the quantities those relationships depend on (speed, draft, displacement). A small number of vessel-specific parameters are then fitted to available data. This approach is intended to produce reliable, physically consistent estimates even for vessels with limited operational history, by fitting to the individual vessel's own fuel reports where available, or, where not, inferring parameters from sister vessels of similar engine type, builder, ship type, and DWT.

The model separately accounts for the main engine (which typically accounts for the large majority of fuel burned underway) and auxiliary engines (generators and boilers), which are modelled as a constant average consumption rate while the vessel is active.

7.3 Coverage & Validation

The consumption model covers just over 46,000 unique IMO numbers across the commercial fleet. Outputs have been validated internally against client operational data as a benchmark, with Bulk Carriers and Tankers showing the strongest validation results to date. Benchmark comparison data was available for a minority of covered vessels, underlining that this is an evolving validation exercise rather than a one-off certification, coverage and validation both continue to expand as more operational data becomes available.

7.4 Best Use & Limitations

• Best suited to: voyage P&L estimation where a client lacks reliable first-party consumption data for a vessel; like-for-like vessel comparison; and the Dry Bulk, Tanker, and Gas Carrier segments, which currently have the strongest model coverage and validation.

• Limited validation: Container, General Cargo, RO-RO, and Reefer vessel types currently have little or no validated benchmark data; VI consumption figures for these types should be used with additional caution.

• MDO profiles: MDO curves are intended to reflect main engine consumption when MDO substitutes for HFO/LFO, but available benchmark data for MDO can sometimes reflect auxiliary engine consumption instead. MDO curves should therefore be treated with additional caution.

• Voyage-average comparisons: because the model reflects the (cubic) relationship between speed and consumption at each point along a voyage, real-world voyage-average consumption, which reflects continuous speed changes due to weather, traffic, and port congestion, will typically read higher than the model's prediction for the same average speed. VI consumption data is not intended to substitute for verified operational data where a client already holds it.

8. Definitions

https://help.veson.com/vesselsvalue/glossary

8.1 Ship Sizes

Standard ship size ranges by deadweight tonnage (DWT), by segment:

Dry (Bulkers)

Wet (Tankers)