Knowledge / Shipyards and production

Ship commissioning: harbour tests, sea trials and acceptance evidence

Connect ship commissioning requirements, safe tests, measurement chains, sea conditions and traceable acceptance evidence.

On this page

Commissioning turns an installed collection of equipment into a demonstrated working ship. The important word is demonstrated: equipment being present, powered or apparently responsive is not the same as evidence that the required function works in the intended configuration. Harbour tests and sea trials contribute different pieces of that evidence. A strong programme connects each requirement to a safe, repeatable test and an accountable acceptance decision, while preserving failures and unresolved limitations rather than hiding them in a completion percentage.

Start with requirements and the tested configuration

Before choosing a test, define what must be shown. A requirement may concern capacity, response time, protection, redundancy, environmental performance or an operating procedure. It should identify the relevant conditions and acceptance basis. “System works” is too vague to test reliably. “The identified function remains available after the specified single failure in the approved configuration” creates a much clearer question, although the actual failure and safe test method still need engineering definition.

Record the configuration: equipment identity, software and parameter versions, valve and breaker states, temporary connections and relevant outstanding defects. A successful result belongs to that tested configuration. If later changes affect the function, determine which evidence remains valid and which tests need repeating. Commissioning records without configuration control can become a collection of true observations that no longer describe the delivered ship.

Distinguish verification, validation and acceptance

Verification asks whether defined requirements have been met. Validation asks whether the system is suitable for the intended use and operating context. Acceptance is a decision by the authorised party under the applicable contract or regulatory arrangement. These concepts overlap in a programme, but a test report should not casually claim all three. A factory demonstration may verify a component requirement without validating the integrated shipboard service.

NASA's product-verification guidance provides a useful general distinction and emphasises traceable evidence, configuration and discrepancies. It is cited as systems-engineering guidance, not a ship-certification rule. In a ship project, owner, yard, suppliers, class and flag representatives can have different roles. Their witness or approval responsibilities must be identified for each test rather than inferred from attendance at the trial.

Build evidence in stages

Factory tests can establish selected equipment properties before delivery. Installation checks establish identity, completeness and specified installation features. Harbour tests can demonstrate functions that do not require unrestricted vessel movement or the full sea environment. Sea trials address the agreed vessel-level performance and behaviour under defined conditions. The exact division varies with ship type, system and contract; there is no universal list that fits every project.

A later stage does not automatically compensate for missing earlier evidence. Discovering a basic wiring or cleanliness problem at sea can be more difficult and hazardous than finding it at the berth. Conversely, a successful berth test may not represent full thermal load, ship motion or integrated demand. The programme should state what each stage establishes and which assumptions remain to be tested later.

Readiness is an engineering gate

Before a test begins, confirm the approved procedure, competent personnel, communications, instruments, safe boundaries, expected response and criteria for stopping. Identify how the system will be restored if the result is abnormal. Temporary test arrangements need the same attention to pressure, electrical safety, isolation and access as permanent arrangements. An improvised workaround can change the very function the test is supposed to assess.

Readiness also includes the condition of neighbouring systems and work teams. A propulsion test may depend on cooling, lubrication, fuel, electrical supply and steering availability. A protective trip test may affect services outside the immediately visible boundary. The responsible team must evaluate these dependencies and obtain the required authorisation. This article does not provide permission to inject faults or defeat protective functions on an operational vessel.

Design the measurement chain

Choose instruments and sampling that match the phenomenon. A steady temperature test and a short electrical transient need different time resolution. Time stamps should be aligned when events from several systems are compared. Calibration, range, resolution, sensor location and data filtering all affect the interpretation. A smooth trend on a display may reflect averaging rather than a physically smooth process.

For an invented example, imagine a 0.2-second pressure excursion and a logger that records one instantaneous sample every second. Depending on timing, the event could occur entirely between samples. Absence from the log would not prove that it never happened. The example does not prescribe a sampling frequency; it demonstrates why measurement design must follow the event duration and required evidence. Sensor response and filtering would also need consideration.

An original performance calculation

Assume a fictional electrical test records a constant additional load of 1.2 MW for 20 seconds. Under that deliberately simple assumption, additional energy is 1.2 × 20 = 24 MJ, or 6.67 kWh. The energy total is small enough to sound unremarkable, but it says nothing by itself about voltage or frequency excursions, protective trips, load sharing or the ability of the generating plant to accept the step.

Now suppose a report contains only average power over a five-minute period. Spreading the same 24 MJ over 300 seconds gives 80 kW. The average describes the energy over that interval, while the original 1.2 MW describes the short demand. Both numbers can be correct and support different conclusions. A commissioning report should preserve the time scale relevant to the requirement rather than replace a transient with a convenient average.

The numbers are authored teaching data, not acceptance limits or a test instruction. A real load-step test requires an approved method, suitable equipment, operating limits and responsible supervision. Its results should include the relevant dynamic response and system state, not just an energy calculation.

Sea conditions belong in the result

Sea-trial performance is influenced by loading condition, wind, waves, current, water depth and other environmental factors. Record the conditions and use the agreed analysis method for the specific test. A measured speed cannot be compared fairly with a contractual target if the displacement, power definition or corrections are inconsistent. The raw observation and corrected result should remain distinguishable.

The ITTC 2024 speed/power-trial procedure, Revision 08, is an authoritative recommended methodology for that particular trial family. It is not a general procedure for every commissioning test. Likewise, the public ISO 19019:2005 scope page describes planning, conducting and reporting sea trials; only its public scope is cited here, not the purchased full standard. Applicable contractual and statutory methods must be established for the actual ship.

Test integrated behaviour and degraded states carefully

A ship can contain individually satisfactory components whose combination behaves poorly. Examples include incompatible control priorities, unintended restart sequences or a shared supply that removes nominal redundancy. Integrated testing should therefore examine defined interfaces and operating modes. The selected tests must be justified by requirements and risk assessment, with safe methods for representing abnormal conditions where direct testing would be inappropriate.

Do not assume that “failure tested” means every failure was tested. Identify the exact injected or simulated condition, the system boundary and the expected recovery. A simulated input may verify control logic while excluding the physical sensor or actuator. A real component isolation may test another part of the chain. The report should say which links were exercised and which remained represented by assumptions.

Regulatory commissioning can have a specific meaning

Some systems have commissioning tests with defined regulatory purposes beyond ordinary supplier startup. DNV's ballast-water management FAQ describes commissioning-test arrangements and accepted testing organisations in that context. This illustrates why the word “commissioned” should not be used without stating the relevant test and authority. A successful self-test or manufacturer's attendance does not necessarily establish completion of every required statutory step.

Check the applicable convention, flag instructions, class requirements and system approval conditions. Product type approval, installation verification, commissioning testing and ongoing compliance are different evidence layers. A certificate at one layer cannot be assumed to cover another without reviewing its scope. The general commissioning programme should make these obligations visible rather than leave them to a final paperwork search.

Preserve anomalies and close the loop

An unexpected result should be recorded with the test state, observed behaviour and immediate response. Diagnose it before repeating the test, and preserve the original data. If parameters or hardware are changed, identify the change and assess its wider effects. Repeating until a favourable run appears, while discarding earlier failures, produces weak evidence and can conceal intermittent problems.

A punch list should distinguish items that prevent safe testing or operation from items that can be closed later under an authorised arrangement. Each needs an owner, due condition and closure evidence. Administrative closure is not the same as technical resolution. Where acceptance with a limitation is permitted, the limitation and responsible approval should be explicit and carried into the operating documentation.

A credible handover

The final evidence package should connect requirements, procedures, tested configurations, calibrated measurements, environmental conditions, results, deviations and authorised conclusions. Crew familiarisation and usable operating information matter alongside test certificates. A ship that passed a function once still needs people who understand its normal modes, limits, alarms and restoration arrangements.

Common mistakes include treating factory tests as proof of complete integration, calling a powered cabinet commissioned, comparing uncorrected and corrected sea-trial values, and losing configuration changes after a successful test. This guide does not define a vessel's complete acceptance programme. Its practical standard is traceability: another competent reviewer should be able to see what was demonstrated, under which conditions, and what remains outside the evidence.

Sources