Writing Commissioning Test Procedures: Structure, Hold Points and Sign-Off

A commissioning test procedure (CTP) is the controlled, step-by-step document that tells the commissioning team how to prove a system or piece of equipment works as designed, and records that it did. A good CTP states its prerequisites, the isolations and safety controls, each test step with an expected result, the hold points where testing stops for sign-off, the acceptance criteria and who signs. If a competent engineer who has never seen the system can run it safely and reach the same verdict as its author, the procedure is doing its job.

This guide covers what goes into a CTP, how to structure the steps, where hold points belong and how to control revisions once testing starts. It is specific to commissioning. For general procedure writing, see WorkProcedures.

What is a commissioning test procedure?

Projects use several names for broadly the same document: commissioning procedure, CTP, functional test procedure (FTP), system test procedure, or commissioning method statement. Some contracts split them. The method statement covers how the work is done safely (resources, access, risk controls). The test procedure covers what is tested and what counts as a pass. Others combine both in one document. Check your contract and the client's commissioning specification before you settle on a template.

What sets a CTP apart from an ITR or checksheet is scope and sequence. An ITR records a check on a single tag or item, such as a motor insulation resistance test or a loop check. A CTP usually covers a system or subsystem and runs in a set order: start the lube oil pump, confirm pressure, prove the low-pressure trip, then permit the main motor start. It sits above the individual records. If you want the full picture of how those documents relate, see our guide to ITPs, ITRs and checksheets.

Where CTPs sit in the commissioning sequence

CTPs are normally run once a system has reached mechanical completion and its pre-commissioning (the 'B' ITRs) is done. They cover energised and dynamic testing: function tests, cause and effect testing, run tests, performance tests and, in some cases, the first introduction of process fluids. The commissioning plan should list which systems need a CTP, who writes each one and when it has to be approved.

The structure of a commissioning test procedure

Most CTPs follow a similar layout, whatever the industry. The table below is a sensible default.

Prerequisites: the section that saves the most time

Most failed test attempts don't fail on the test. They fail because something that should have been finished wasn't. A mis-set trip, a missing blind or an instrument still in its preservation bag will stop the day. List the prerequisites as checkable items, not prose:

• Mechanical completion certificate issued for the system, or the subsystems in scope
• All 'A' and relevant 'B' ITRs approved, including loop checks and motor solo runs
• No open Category A punch items against the system (terminology varies, so use your project's categories)
• Relief valves installed, set and certified
• Instruments in scope calibrated, with certificates in date
• Control system software at the approved revision, with forces logged
• Utilities available: power, instrument air, cooling water, nitrogen as needed
• Permits and isolations in place, and the area handed to commissioning control

The test lead should confirm each prerequisite and initial it before step one. If a prerequisite is waived, record who approved the waiver and why.

Safety and isolations

Keep this section specific to the test rather than repeating the site safety rules. Name the energy sources involved, the isolation points (by tag), the permits required and anything unusual. Examples include a temporary strainer, a vent to a safe location, or a system that shares a header with plant already in service. Where the test crosses the construction-to-commissioning boundary, say which permit regime applies.

Writing test steps that can be followed

Each step should be one action with one observable result. A practical format:

The values above are illustrative. Take real setpoints from the cause and effect chart or instrument index, not from memory. Some rules that stop arguments later:

• State the expected result before the test, not after. "Observe pump behaviour" can't fail, so it proves nothing.
• Quote the source of every limit (datasheet, vendor manual, design basis). If a witness challenges a value, you can point to it.
• Record the actual value, not just a tick. "1.49 barg" is evidence; a tick isn't.
• Keep one verification per step. If step 5.4 has three things to check, split it or give each its own result cell.
• Include the restoration steps. Forces left in a safety system after testing are a well-known failure mode.

Hold points in a commissioning procedure

A hold point is a step where testing stops and doesn't continue until a named person has reviewed or witnessed the result and released it. They belong in the CTP itself, marked clearly, not left to verbal agreement on the day.

Typical places for a hold point:

1. Prerequisite confirmation, before any energisation or introduction of fluids
2. First energisation of a motor, heater or drive
3. Before and after proving a safety function, such as a trip or shutdown valve closure
4. Introduction of hydrocarbons, toxic or flammable media, or first steam
5. Before a performance or run test, where the client or vendor must witness
6. On completion, before restoration removes the evidence

Witness points differ from hold points. Work may proceed if the notified witness does not attend. Your ITP or contract should define both, along with notice periods. Our guide to hold points and witness points covers release authority and missed witnesses in detail.

Acceptance criteria and dealing with failures

Acceptance criteria should be measurable and agreed before the test, ideally when the procedure is approved. "Pump runs satisfactorily" invites a debate. "Discharge pressure within ±5% of the datasheet duty point at rated flow, bearing temperature stable below the vendor limit for a two-hour run" does not. Again, take the numbers from the vendor or design documents.

When a step fails, the procedure should say what happens next. Typically:

• Stop at the failed step and make the system safe
• Record the actual result and raise a punch item or deficiency against the tag
• Decide with the responsible engineer whether the fix needs a retest of that step only, a section, or the whole procedure
• Record the retest on the same procedure revision, with a reference to the punch item

Don't correct a failed result on the record sheet by overwriting it. The failure and the retest are both part of the evidence.

Review, approval and revision control

A CTP normally goes through an internal review (commissioning, operations, process or electrical engineering as relevant, and HSE), then to the client for approval if the contract requires it. Vendors should review the steps that touch their equipment. Their manuals often contain start-up conditions that the design documents don't.

Revision control matters most once testing has started. Common problems:

• Two revisions in circulation, one on the field copy and a newer one in the document system
• Setpoints changed after a management of change, but the CTP not updated
• Red-line changes made in the field and never incorporated

Set a simple rule. Only the current approved revision is used for execution. Any field change is recorded as a formal deviation, approved by the named authority, and folded into the next revision. The executed copy, with actual results and signatures, becomes a handover record in the system dossier.

Sign-off: who signs and what it means

Sign-off blocks should say what each signature confirms. A typical set:

• Test lead: the steps were performed as written and the results are recorded accurately
• Discipline engineer: the results meet the acceptance criteria
• Vendor representative (where required): the equipment was operated within its limits
• Client or operations witness: witnessed and accepted, with any exceptions noted

The completed CTP, together with the approved ITRs and the punch status, feeds the certificate that closes the commissioning stage for that system. On many projects that is a ready-for-start-up or equivalent certificate; see ready for start-up (RFSU) for what that stage certifies.

Commissioning test procedure checklist

Before you issue a CTP for approval, check:

• System and subsystem numbers match the current systemisation
• All references are at the current revision
• Prerequisites are listed as checkable items, including certificates and punch status
• Isolations and permits are named by tag and type
• Every step has one action and one expected result, with a source for each limit
• Hold points are marked, with the release authority named
• Failure handling and retest rules are stated
• Restoration steps remove all forces, overrides and temporary equipment
• Sign-off blocks say what each signature confirms

Managing CTPs in Accord

Accord includes a Commissioning Test Procedures module for authoring and revisioning CTPs, enforcing hold points and executing them step by step, with completed procedures feeding ITR closure. Because CTPs sit in the same system hierarchy as ITRs, punch items and certificates, the dependency engine stops 'B' records starting before their 'A' prerequisites are approved, and certificates can't be issued without their prerequisites. For site-wide operating and maintenance procedures outside commissioning, WorkProcedures generates structured SOPs with revision history and staff acknowledgement tracking.

Frequently asked questions

What is the difference between a commissioning procedure and a method statement?

A method statement mainly describes how work will be carried out safely: sequence, resources, access and risk controls. A commissioning test procedure defines what is tested, the expected results and the acceptance criteria, and records the actual results. Some projects combine them in one document, so check your contract's requirements.

Who writes commissioning test procedures?

Usually the commissioning engineer for the system, with input from the discipline engineers, operations and the equipment vendor. The client or operator often reviews and approves them, particularly for safety functions and performance tests.

How detailed should a functional test procedure be?

Detailed enough that a competent engineer unfamiliar with the system could run it safely and reach the same pass/fail result. That normally means individual numbered steps, an expected result for each, and acceptance limits quoted from a named source.

Can you change a CTP during testing?

Only under control. Field changes should be recorded as a deviation, approved by the named authority before the step is performed, and incorporated in the next revision. Uncontrolled red-lines are a common source of disputes at handover.

What happens if a commissioning test fails?

Stop, make the system safe, record the actual result and raise a punch item or deficiency. Once the fault is corrected, retest the affected steps as agreed with the responsible engineer, and keep both the failed and passed results in the record.