The Class-Society Evidence Package

A class society approves a safety system on one question: can you defend every claim with reproducible test data, signed by an engineer whose name appears on the package? Marketing language has no standing. Where the sibling notation suites decide which capability you can earn, the evidence package is the dossier that earns it — and the packaging discipline is more important than most teams realise.

What the package contains

At minimum, a submission a surveyor can act on has five parts, each traceable to the next:

• Functional description with every claim traced to a specific, numbered test case.
• Test methodology — the abuse-rig setup, the calibration of the heat/off-gas source, and the pass/fail criteria fixed before the run.
• Raw and processed data, with unit, timestamp and sample-rate discipline so a third party can re-derive the result.
• Algorithm documentation — the detection logic's inputs, parameter ranges and declared failure modes, at the level a reviewer needs to judge behaviour (not the proprietary internals).
• Reliability evidence — environmental qualification, EMC, ingress rating and network fault behaviour.

The standards the tests have to map to

A class reviewer does not evaluate a detection claim in isolation — they check it against the recognised test baselines for shipboard equipment. Marine electronics are expected to demonstrate environmental and EMC endurance to IEC 60945 (the general requirements for shipboard navigation and radio-communication equipment), an ingress rating stated to IEC 60529 (the IP code), and, for the fire-detection function itself, behaviour consistent with the fire-detection and alarm expectations under SOLAS Chapter II-2 and the FSS Code. A package that presents its results already mapped to these references — rather than as bespoke bench numbers — is one a surveyor can accept without re-scoping the test.

Why methodology beats results

A number a reviewer cannot reproduce is worse than no number, because it invites doubt about everything around it. The methodology section is where approval is won or lost: the source has to be calibrated and traceable, the trip criteria fixed in advance, and the run repeatable on demand — ideally under witness. For a lithium-ion detection claim, that increasingly means abuse profiles anchored to a recognised cell-level method such as UL 9540A, so the staged event the detector is tested against resembles the runaway it will meet at sea rather than a convenient bench proxy.

Reliability and failure-mode evidence

Detection performance is only half the dossier; the other half is what the system does when part of it fails. Class reviewers expect a documented failure-mode analysis (FMEA) covering sensor loss, segment loss and power loss, with the alarm behaviour in each state stated and tested — a fault must annunciate, not fail silent. Mean-time-between-failure data, the environmental qualification envelope, and the network's behaviour under a severed or degraded link belong here. Type approval turns on this section as much as on the detection curves.

What it does not contain

• Claims about competing systems, even comparative.
• Marketing material of any kind.
• Benchmarks the auditor cannot re-run on request.

Why the packaging discipline pays off

The same dossier that earns type approval shortens every downstream step: a witnessed-trial programme, a notation survey, and the underwriter questionnaire that increasingly asks for auditable evidence rather than assurances. A detection layer whose data architecture already produces signed, retention-managed, reproducible records is one where the package is a by-product of how the system runs — not a document assembled under deadline after the fact.

Sources

• IEC 60945 — Maritime navigation and radiocommunication equipment and systems: general requirements, methods of testing and required test results (environmental / EMC baseline for shipboard electronics).
• IEC 60529 — Degrees of protection provided by enclosures (IP code).
• IMO SOLAS Chapter II-2 and the FSS Code — fire-detection and alarm system requirements against which a marine detection function is assessed.
• UL 9540A — test method for evaluating thermal-runaway fire propagation in battery energy-storage systems (cell-level abuse reference for staged detection tests).
• IACS Unified Requirements and member type-approval schemes (e.g. DNV product certification) — the framework under which a product certificate / type approval is issued.
• [VERIFY: exact clause and current edition of each standard shift over time — cite the edition in force at submission; society-specific document numbers vary by class.]