What Should a Control System Site Acceptance Test Cover?
A control system site acceptance test should demonstrate that the installed system works with the actual equipment, field interfaces and site services within an agreed scope. Your checklist should cover signal checks, utilities, operating sequences, alarms, failure recovery and recorded results witnessed by the responsible people. Factory testing provides earlier evidence of agreed functionality; site testing must establish what works after installation and identify anything still unproven.
How does site acceptance testing differ from factory testing?
Use factory acceptance testing to check the assembled system against the agreed design before delivery, using simulation where site equipment is unavailable. Use site acceptance testing to check the installed arrangement, including actual wiring, instruments, drives, valves, networks and connections to other systems. For open defects after testing, see how to manage a commissioning punch list.
Keep those evidence boundaries visible. A simulated valve response during factory testing does not establish that the installed valve moves correctly. A controller receiving a simulated temperature does not establish that the field instrument has the correct range or appears under the correct equipment name.
Carry the factory test record into the site plan. Identify which results remain relevant, which require confirmation after installation and which tests could only be completed on site.
What should be agreed before testing starts?
Write acceptance criteria before the witness session. Each test needs a requirement reference, starting conditions, actions, expected results and a place for observed results.
Use the approved functional design specification as the starting point. Resolve gaps before testing: the witness session should assess agreed behaviour, rather than become a debate about what the system was meant to do.
Agree:
- The equipment, operating modes and interfaces included.
- The permitted test conditions and available utilities.
- Who operates equipment, witnesses results and authorises acceptance.
- Which checks require specialist participation.
- How defects, unavailable tests and retests will be recorded.
- What prevents acceptance and what may remain open under stated conditions.
Plan the testing window alongside isolation, installation and restart work. Our guide to planning a PLC retrofit around a shutdown window covers those dependencies.
What belongs in the site acceptance test checklist?
Use this control system site acceptance test checklist as a planning framework. Expand each row into test steps for the actual plant.
| Area | What to demonstrate | Evidence to retain |
|---|---|---|
| Installed configuration | Equipment identities, software versions and settings match the approved installation | Configuration record and deviations |
| Field interfaces | Signals and commands reach the correct physical equipment | Signal checks with observed responses |
| Site services | Required supplies are available and their loss produces agreed behaviour | Supply conditions and recovery results |
| Operating sequences | Starts, stops, transitions and permissives follow the design | Witnessed sequence records |
| Alarms and faults | Conditions appear clearly and require the agreed response | Alarm text, state and acknowledgement results |
| External interfaces | Data and commands pass correctly between connected systems | Observed values and communication-loss results |
| Recovery | Restarts and reconnections follow the agreed rules | Recovery observations |
| Handover | Records, backups, instructions and open items are complete | Handover register and acceptance decision |
Do not treat a tick against an entire row as sufficient evidence. Record the individual checks beneath it.
How should field interfaces and utilities be checked?
For each instrument, follow the signal from the field device through the controller to the operator display. Check identity, range, units, scaling and the response to an unavailable or invalid signal.
For each output, confirm which physical device responds. Where movement is authorised, observe direction, travel, feedback and the relationship between command and status. Check local and remote selection and any permissions that govern control.
Record whether the test used the actual device, an injected signal or a simulation. State the remaining boundary whenever the complete path was not exercised.
Include the services that the installation depends on: electrical supplies, instrument air, cooling water, communications or other relevant utilities. Establish which conditions are available for testing and how interruptions may be introduced safely.
Do not disconnect a live service simply to complete a checklist. Agree the method with site operations and the responsible specialists, then record any condition that could not be tested.
Which operating sequences and failure cases should be witnessed?
Begin with normal operation, then test the branches that determine whether the system behaves correctly when conditions change.
Include:
- Start permission present and absent.
- Normal start, controlled stop and interrupted operation.
- Manual and automatic mode changes.
- Equipment unavailable during a sequence.
- Missing feedback or a response outside the agreed time.
- Loss and restoration of a relevant utility or connection.
- Recovery after a controller or operator station restart.
- Operator actions needed before operation can resume.
Observe the physical response alongside the screen indication. Record whether the system waits, stops, inhibits a command or requests intervention, as specified.
For alarms, check the message, equipment identity, operator response and return to normal. Acknowledgement and clearance should be tested separately where the design distinguishes them. See why operators ignore alarms when reviewing whether messages give useful direction.
How should safety-related functions be handled?
Agree safety-related test scope and methods with a qualified assessor and the people responsible for site safety. Identify who may authorise testing, what precautions apply and what evidence is required.
Keep routine control checks and safety-related acceptance records clearly identified. Describe exactly what was exercised and who witnessed it.
Where a safety-related test remains incomplete, refer acceptance of that function to the responsible qualified people. Do not replace the missing evidence with a general statement that commissioning is complete.
What does a useful witnessed result look like?
A useful record connects the requirement, action and observation closely enough for someone else to understand the result.
Illustrative example: A transfer pump may start only when the receiving vessel is available and the route valve confirms open.
The test record could contain:
- Starting condition: Receiving vessel available; route valve closed.
- Action: Operator requests transfer.
- Expected result: Valve receives an open command; pump remains stopped until open feedback arrives.
- Observed result: Record actual valve movement, feedback, pump behaviour and displayed messages.
- Failure branch: Withhold open feedback using the approved method; check the specified timeout and operator indication.
- Recovery: Restore feedback and verify the specified restart action.
Add the outcome, witness, configuration reference and any defect reference. Preserve the original failure alongside the retest result.
Where should the acceptance boundary be drawn?
Separate completion of a test from acceptance of the whole installation. Define which unresolved items prevent operation, which prevent final acceptance and which may remain open with an owner and closure condition.
Mark unavailable tests as not tested. If product, a utility or a connected system is unavailable, state the limitation and arrange the remaining demonstration.
Also distinguish functional acceptance from any separate production trial. If sustained output, product quality or operation across a wider process range matters to acceptance, give it a defined test scope and pass criteria.
The handover record should identify accepted functions, restrictions, open defects, outstanding tests and responsibility for closure. Include the configuration and backup associated with the accepted results.
What common mistakes weaken a site acceptance test?
Repeating factory tests without checking installed interfaces. Retain useful earlier evidence, but test the physical connections and site dependencies.
Recording only “pass”. Capture the response that supports the decision, especially for faults and recovery.
Using simulations without identifying them. Record what was substituted and what remains unverified.
Changing settings without revisiting affected tests. Assess which earlier results need retesting after each change.
Signing with vague open items. Give each item an owner, consequence and closure condition.
Treating an operator demonstration as the entire test. Check abnormal conditions and evidence requirements as well as routine operation.
Related reading: control system handover files and access and defining manual and automatic machine modes.
Frequently asked questions
Can site testing begin before every connected system is ready?
Agree a staged scope if necessary. Record unavailable interfaces explicitly and schedule their remaining tests before accepting those functions.
Who should witness the tests?
Select witnesses by responsibility. Include operations, maintenance, the integrator and relevant equipment or specialist representatives where their acceptance is needed.
Does every factory test need repeating on site?
Decide test by test. Repeat checks affected by installation, configuration changes or actual equipment connections, and document the basis for retaining earlier evidence.
What should happen when a test fails?
Record the observation, assess its effect on acceptance, correct the cause and retest the affected behaviour. Keep both results.
Preparing for commissioning or a retrofit? Contact us to discuss the installed interfaces, test conditions and acceptance evidence your project needs.