A factory acceptance test is not a ceremony for signing a certificate. It is the last point before shipment where the machine can be opened, instrumented and run without a building load. If the test is witnessed properly, it settles the capacity, efficiency, pressure integrity and control questions while the machine is still in the factory.
This article sets out a witness plan for a 400 kW water-cooled screw chiller. The tolerances are project specifications; the contract and the manufacturer’s test procedure take precedence.
Four tests, four questions
| Test | Question it answers | Witness point | Record |
|---|---|---|---|
| Capacity | Does it deliver 400 kW at the specified temperatures? | 100%, 75%, 50% and 25% load | Water flow, temperatures, power |
| Efficiency | Is the COP at least the quoted value? | Same points as capacity | Power meter, calibration certificate |
| Pressure integrity | Does the refrigerant circuit hold pressure? | After assembly, before charge | Pressure, time, temperature |
| Controls and safety | Do the sequences and interlocks work? | Simulated faults | Alarm log, screen captures |
Capacity test
The test should run at a steady state for at least 30 minutes at each point, after the temperatures and power have stabilized. A 400 kW chiller at 7/12 °C has a design flow of 400 ÷ (1.163 × 5) = 68.8 m³/h. The condenser flow depends on the heat rejection and the specified rise.
Measure the chilled-water flow with a calibrated meter, the entering and leaving temperatures with matched sensors, and the compressor power with a class 0.5 meter. Calculate capacity as flow × density × specific heat × temperature difference. Do not accept a capacity calculated from the compressor displacement or from a nominal curve.
A capacity tolerance of ±5% is common in a project specification. A shortfall at 100% load may be a test-condition issue, a flow issue or a real machine problem. Check the water flow and the temperature sensors before accepting a derating.
Efficiency test
COP is capacity divided by total input power, including the compressor, the oil pump and the control panel, but excluding the chilled-water and condenser-water pumps unless the contract says otherwise. State the boundary in the test report; a COP that excludes the oil pump is not comparable with one that includes it.
Take at least four points: 100%, 75%, 50% and 25%. A single full-load COP does not show the part-load behavior. The IPLV is a weighted figure; it is useful, but it is not a substitute for the measured points.
Pressure and leak test
The refrigerant circuit should be pressure-tested before charging. A common specification is 1.5 times the design pressure, held for 30 minutes, with no visible drop after temperature correction. For a system with a design pressure of 2.5 MPa, the test is 3.75 MPa. Do not use the operating pressure as the test pressure.
Record the ambient temperature, the starting pressure, the ending pressure and the hold time. A pressure drop can be a leak or a temperature change; correct to a common temperature before calling it a failure. A helium leak test is more sensitive and is worth requiring for a large machine.
After the pressure test, evacuate the circuit and record the vacuum. A standing vacuum test can reveal moisture or a small leak. The manufacturer’s procedure should state the vacuum level and the hold time.
Controls and safety
Do not leave the control test to the site. Simulate a chilled-water flow failure, a condenser high-pressure trip, a low-pressure trip, a sensor failure and a communication loss. Confirm that the compressor stops, the alarm is latched and the restart requires a manual reset where the specification requires it.
Check the start sequence, the pump interlock, the tower enable signal and the BAS interface. If the machine is supplied with a gateway, test the points that the client will actually use, not only the vendor’s demo page. Record the point list and the alarm text.
Instrumentation
Ask for the instrument list before the test. Flow, temperature and power should be measured with calibrated instruments, and the calibration certificates should be current. A test that uses the machine’s own display for capacity is not independent; the controller may be reading a sensor that is part of the fault.
Record the sampling interval and the averaging period. A single snapshot at 100% load is not a steady-state test. The log should show the temperatures and power over the hold period, so the reviewer can see whether the machine was stable or drifting.
Noise and vibration
Measure sound pressure at 1 m from the machine at the specified operating point. A 400 kW screw chiller may be 75–85 dB(A); the project limit may be lower. Vibration should be within the manufacturer’s limit at the bearings and the frame. A loose panel or a pipe support can dominate the reading; check the installation before blaming the compressor.
The witness file
Bring a checklist and a camera. Record the serial number, the model, the software version and the instrument calibration dates. Photograph the nameplate, the pressure gauge, the power meter and the alarm screen. Ask for the raw data file, not a summary curve. A test report without raw data cannot be rechecked.
If the machine fails a point, do not sign “subject to site retest” unless the contract allows it. A factory retest is cheaper than a site dispute. The whole purpose of the FAT is to find the problem before the machine leaves the factory.
Common questions
What is a chiller FAT?
A factory acceptance test is a witnessed test of capacity, efficiency, pressure integrity and controls before shipment. It is the last chance to correct a machine problem without site costs.
How long should a capacity test run?
At least 30 minutes at each load point after temperatures and power stabilize. Four points (100%, 75%, 50%, 25%) are typical for a project specification.
What tolerance is normal for capacity?
Plus or minus 5% is common, but the contract governs. A shortfall should be checked against water flow, temperature sensors and test conditions before acceptance.
Does COP include the pumps?
State the boundary. A chiller COP usually includes the compressor, oil pump and panel, and excludes the chilled-water and condenser-water pumps. A COP with a different boundary is not comparable.
How is the pressure test performed?
At 1.5 times the design pressure for 30 minutes, with temperature correction. Record the start and end pressure, and use a helium test if a smaller leak must be detected.
Should controls be tested at the factory?
Yes. Simulate flow failure, high and low pressure, sensor failure and communication loss. Confirm the stop, alarm and reset behavior before shipment.
What tests to witness?
Capacity, power, sound and control response. Watch the instruments, not the printout. A test you did not see is a claim.
What if a test fails?
Stop acceptance and record the deviation. A retest plan with a date is better than a verbal promise.
Further technical reading
For manufacturer due diligence, see Chiller Manufacturer Audit. For site acceptance, see Chiller Acceptance: Standards, Tests and Files. For commissioning, see Water Chiller Commissioning. For installation, see Water Chiller Installation: mechanical room Layout.
Related reading
Chiller Acceptance Standards · Chiller Factory Acceptance Test · Chiller Tender Technical Response