An annual chiller shutdown is often written as a list of tasks: clean the tubes, change the oil, check the electrical, start the machine. A list is useful, but it does not tell the technician what failure each task prevents. This article organizes the shutdown by failure mode, so the work has a purpose and the acceptance criteria are clear.
The four families are water side, refrigerant side, electrical and controls. Each has a set of inspections that should produce a number, not a tick.
Before the shutdown
Record the baseline: chilled-water in/out, condenser-water in/out, currents, pressures, approach, oil level, purge count and alarm history. A shutdown without a baseline cannot prove that the service restored performance. Photograph the sight glass, the tube ends and the electrical panel before opening anything.
Lock out the power and the water. Isolate the machine, tag the valves and drain the sides that will be opened. A chiller is a pressure vessel and an electrical machine; the lockout is not a formality.
Water side: the first family
| Task | Failure it prevents | Acceptance |
|---|---|---|
| Tube brushing / mechanical cleaning | Scale, biofilm, reduced heat transfer | Tubes clear, approach restored |
| Eddy-current test | Tube wall loss, pinhole leak | No tube below 80% wall |
| Strainer and water treatment | Solids, corrosion, Legionella | Clean strainer, chemistry in range |
| Flow check | Low flow, high approach | Design flow ±10% |
A 0.5 mm scale layer can cut heat transfer by 10–20% and raise the condensing pressure. A tube cleaning that restores the approach can recover 10–20% of the energy, which is usually the largest single saving in the shutdown. The eddy-current test is the only way to find a tube that is about to leak; a visual inspection sees the inlet, not the wall.
On the evaporator side, the water is clean but the tubes can still foul with corrosion products. Clean and test both bundles. If the machine has a flooded evaporator, check the oil return port while the bundle is open.
Refrigerant side: the second family
Check for leaks at the joints, the valve packing, the relief valve and the purge. An electronic leak detector is better than soap bubbles for a small leak; a pressure hold test is better for a large one. Do not add refrigerant to a machine with an unknown leak; find the leak first.
Check the charge against the nameplate and the subcooling. A machine that is low on charge may still run, but the approach and the current will be wrong. A machine that is overcharged may have a high condensing pressure and a flooded condenser.
For a screw, change the oil filter if the ΔP is above 0.1 MPa or at the manufacturer’s interval. For a centrifugal, check the purge count and the oil level. Oil analysis is the best evidence of internal wear: water, acidity, viscosity, iron, copper and refrigerant dilution. A sample taken before the oil change is the baseline for the next year.
Electrical: the third family
Measure the insulation resistance of the compressor motor and the pump motors. A low reading is a warning of moisture or winding degradation. Check the contactor contacts for pitting, the terminal torque, the overload setting and the phase imbalance. A 2% voltage imbalance can cause a significant current imbalance and a motor hot spot.
Test the safety chain: high-pressure, low-pressure, oil pressure, flow switch, freeze stat and overload. Test each input by the manufacturer’s simulation method, not by blocking a fan and waiting for a trip on a live compressor. Record the set point and the actual trip value.
Check the crankcase heater and the control transformer. A failed crankcase heater allows liquid refrigerant to sit in the oil during a long shutdown; the next start can damage the compressor. A control transformer that is hot may be undersized or overloaded.
Controls: the fourth family
Calibrate the temperature and pressure sensors against a reference. A sensor 2 K off will shift the control and the energy bill. Check the valve actuators, the guide vane linkage and the VFD cooling fan. A guide vane that is mechanically free but mis-calibrated will command the wrong opening.
Back up the controller program and the parameters. A shutdown is the best time to record the set points, the staging sequence and the alarm limits. If the controller battery is old, replace it before it loses the program.
Restart and acceptance
Evacuate and charge if the circuit was opened. Open the water valves slowly to avoid a pressure surge. Start the pump first, prove the flow, then start the chiller. Record the same baseline points as before the shutdown, at the same load if possible. The acceptance is not “it runs”; it is “the approach, current and capacity are at or better than the baseline”.
Run for at least one hour at a stable load. Check for abnormal noise, oil level, purge count and water chemistry. If the machine cannot reach the baseline, the service is not complete.
Boundary
This shutdown procedure is for a water-cooled screw or centrifugal chiller. An air-cooled chiller has no tube bundle but still needs the coil, electrical and refrigerant checks. A chiller with a VFD has additional drive checks; follow the manufacturer’s drive manual.
Common questions
How often should a chiller be shut down for service?
Annually for most comfort-cooling plants. A critical process plant may need a mid-year inspection. The water quality and the running hours decide whether the interval should be shorter.
What is the most valuable annual task?
Tube cleaning and eddy-current testing on the water side. Restoring the approach can recover 10–20% of energy, and the eddy-current test finds a tube before it leaks.
Should I change oil every year?
Not necessarily. Use the oil analysis and the manufacturer’s interval. A clean machine may go 5,000 hours; a wet site may need 2,000. Change the filter when the ΔP is above 0.1 MPa.
How do I test the safety switches?
Use the manufacturer’s simulation mode or a test port. Do not block a fan or a pump and wait for a live trip. Record the actual trip value.
What is a good insulation reading?
Follow the manufacturer’s minimum, often in the hundreds of megohms for a dry motor. A falling trend is more important than a single pass/fail number.
What proves the service is complete?
The baseline is restored: approach, current, capacity and oil/purge readings. A machine that merely starts is not proven.
What are the four failure modes?
Water-side fouling, refrigerant loss, electrical degradation and control drift. Each has a measurement. Inspect by mode, not by a generic checklist.
What acceptance criteria?
Approach, pressure drop, superheat, insulation resistance and control response. Set the limit before the shutdown. A reading without a limit is an opinion.
Further technical reading
For installation hold points, see Chiller Installation: 90 Days, Four Hold Points. For water treatment, see Cooling Water Treatment: Scale, Corrosion and Biofilm. For components, see Chiller’s Four Core Components. For alarm response, see Chiller Alarm: The First 30 Minutes.
Related reading
Evaporator water-side cleaning · Chiller plant sensors · Water chiller sizing