In short. A control system is accepted on evidence, not on belief. FAT proves the logic against a simulated plant; SAT proves the integrated site, including failures. Each test has a written expected result, and the full power-loss sequence is measured with time stamps.
Stages
| Stage | Where | Proves |
|---|---|---|
| Document review | Office | Logic, drawings and cause-and-effect agree |
| FAT | Factory or workshop | Control logic and panels against simulation |
| Installation checks | Site | Wiring, polarity, calibration, insulation, earthing |
| SAT | Site, no load then load | Each function with real equipment |
| Integrated system test | Site | Complete power-loss and recovery sequence |
| Performance run | Site, under load | Stability, sharing, thermal behaviour |
What the plan includes
- Test cases from the cause-and-effect matrix. See cause and effect.
- Each case: precondition, action, expected result, pass or fail, signature.
- Failure injection: lost signal, controller power cycle, network break, sensor failure.
- Safety precautions and permits for live tests.
- Measuring instruments and calibration records.
The integrated test
Simulate utility failure and record T0 to T5: failure, start command, generator running, source acceptable, transferred, load accepted. Then run recharge, restoration and utility return with retransfer. Finish with a failed first start. See transfer architecture and commissioning.
Performance checks
- Load sharing accuracy at several load levels.
- Voltage and frequency dips and recovery for planned steps.
- Load-shed response time.
- Synchronising window and closing angle.
- Alarm and event records match actual events.
Use a resistive load bank for repeatable loading. See load-bank testing.
Sign-off
Close punch items, update drawings and logic to as-built, back up programs, train operators and hand over test records. Define a warranty and a re-test schedule.
Illustrative example
SAT of a three-generator plant finds that load shedding trips the wrong feeder because two cables were swapped in the field. FAT could not detect it. Because the SAT case injects the shedding event and checks the specific breaker, the fault is found and corrected before handover.
Three acceptance stages
| Stage | Where | Purpose |
|---|---|---|
| FAT (factory acceptance test) | At the panel builder | Prove the logic, I/O and interfaces before shipping, with simulators |
| SAT (site acceptance test) | On site with the real equipment | Prove wiring, interfaces and sequences with real breakers and generators |
| Integrated test | Whole system under load | Prove behaviour in realistic scenarios |
Test plan content
- Test cases traced to the functional specification and cause-and-effect chart.
- Expected result and pass criteria for each case.
- Failure and abnormal cases: lost signals, wrong sequences, equipment failure.
- Record sheets with signature, date and observed values.
- Punch list of defects with owners and closure dates.
Essential scenarios
- Mains failure and restoration, with real load.
- Generator fails to start.
- Generator trips on load.
- Loss of control power and of communications.
- Simultaneous events.
- Load shedding and restoration.
Change control after acceptance
Keep the as-tested software version. Any change needs a revision record and a retest scope. See cause-and-effect engineering and commissioning.
The master engineering question
For every function the plant depends on, what recorded test proves that it works, and when was it last repeated?
What to send us
Send scope, logic and cause-and-effect. Send your application details.
Questions engineers ask
Is FAT necessary?
It reduces site risk and time, especially for complex logic.
Who witnesses tests?
The owner or their representative, with the contractor.
Can tests be done at no load?
Many, but load tests are needed for sharing and stability.
How often to retest?
After changes and on a schedule that fits risk.
This guide gives general principles. Tests must follow site safety rules.