The most common acceptance argument sounds like this: the contract says 500 kW, the nameplate says 500 kW, but a site measurement with an ultrasonic flow meter reads 86 m³/h and 4.6 K, which works out to 460 kW. The client calls the machine defective; the supplier says the condition is wrong.
Nine out of ten such disputes are not quality problems. They are a failure to fix the acceptance basis in the technical agreement. Chiller capacity, input power and COP all depend on condition: 12/7 °C chilled water with 30/35 °C condenser water is one set of numbers; 32/37 °C condenser water is another. Capacity can differ by 3–5% and input power by 6–8%. The first acceptance step is therefore not to measure, but to write down which condition, what tolerance, which instruments and who reads them.
Three standard layers, not one
Chiller standards fall into separate layers. Mixing them produces mismatches such as using a product standard to judge installation quality, or a construction code to judge machine performance.
| Layer | Main standard | What it covers | Use at acceptance |
|---|---|---|---|
| Product performance | GB/T 18430.1-2007 | Nominal condition, test method, tolerance | Capacity, input power, COP |
| Product safety | GB 9237-2001; GB/T 150.1-2011 | Refrigeration safety; pressure vessels | Safety valve, burst disc, vessel certificate |
| Energy efficiency | GB 19577-2024 | Efficiency limits and grades | COP, IPLV, energy label |
| Project acceptance | GB 50243-2016 | Work breakdown, inspection lots, records | Installation quality and test-run records |
| Construction | GB 50738-2011 | Pipework, insulation, electrical work | Pressure test, flushing, insulation thickness |
The product standard is valuable because it defines tolerance. GB/T 18430.1-2007 sets a lower limit for measured capacity and an upper allowance for input power; it does not require an exact match. Put that tolerance in the contract and there is a basis for discussion. Leave it out and both sides argue from memory.
GB 19577-2024 is the mandatory efficiency standard. At acceptance, the label grade must match the registration, and measured COP outside the allowed deviation fails the efficiency check. This is sampled in energy audits and green-factory applications, so the label is not a formality.
Documents: four groups and the usual gaps
More documents is not better; every item must be traceable. Group them by issuer, because a missing group blocks a different stage.
| Group | Documents | Issuer | If missing |
|---|---|---|---|
| Random file | Certificate, factory test report, type test report, packing list, manual, wiring diagram | Manufacturer | Cannot register or verify efficiency |
| Compliance | Pressure-vessel certificate, safety-valve calibration, refrigerant charge and recovery record | Manufacturer / inspector | Special-equipment acceptance fails |
| Construction | Foundation, pressure test, flushing, holding pressure, vacuum, insulation and earth tests | Installer | Concealed work cannot be traced |
| Commissioning | Test-run log, control parameter table, spares list, training record | Commissioning party | Maintenance has no baseline |
The two most frequently missing items are the type test report and the control parameter table. Many projects receive only a factory test report, while tenders, efficiency registration and green-factory applications often ask for a type test. The parameter table is almost never handed over, yet it is the maintenance baseline: without it, no one can tell whether a pressure protection value was changed, or by whom.
Pressure vessels deserve attention. Shell-and-tube condensers and evaporators may fall within the pressure-vessel scope, requiring a certificate and local use registration. Missing that paper stops the project at the special-equipment stage, unrelated to chiller performance.
Open-box check: five items, three records
The open-box check fixes the condition on arrival so responsibility is clear later. Check five items: nameplate data (model, nominal capacity, input power, refrigerant type and charge, serial number and date); the technical agreement item by item (condition, voltage, water pressure drop, connection size, dimensions, operating weight); visible transport damage; nitrogen holding pressure (typically 0.02–0.05 MPa; a rush of gas on opening the valve indicates no leak); and the accessory box (rubber pads, anchor bolts, flow switch, temperature sensors, Y-strainer).
Keep three records: a signed open-box sheet, photographs of the nameplate and any damage, and the accessory count. These are the evidence for a later claim. A project that failed to photograph a dented condenser end cover ended up paying for the replacement because the manufacturer and the carrier each blamed the other.
Installation and pressure holding
Foundation level should be 2–3 mm/m in both directions. Select anti-vibration pads by dividing operating weight across four points, and keep static compression at 5–10 mm. An uneven base transmits vibration to the floor and produces noise complaints; it also adds stress at pipe joints, which leaks within a season.
Water-side pressure test: 1.5 times working pressure and not less than 0.6 MPa, hold 10 minutes without leakage, then reduce to working pressure and hold 60 minutes within the allowed drop. Test chilled water and condenser water separately. Testing them in series hides a leak on one side behind make-up on the other.
Refrigerant-side tightness: charge nitrogen to the manufacturer’s pressure, commonly 1.0–1.8 MPa depending on model, and hold 24 hours. Correct the pressure drop for temperature before judging. Absolute pressure is proportional to absolute temperature: p₂ = p₁ × T₂ ÷ T₁, with T in kelvin. If the starting absolute pressure is 1.20 MPa at 20 °C (293 K) and the temperature falls to 12 °C (285 K) after 24 hours, the theoretical pressure is 1.20 × 285 ÷ 293 = 1.167 MPa. A reading of 1.16 MPa is normal; 1.05 MPa indicates a leak. Without the correction, a day-night swing of 8 K produces about 2.7% change and a false alarm.
Evacuate to an absolute pressure no higher than 133 Pa and hold for one hour without rise. Poor vacuum leaves moisture that freezes at the expansion valve and forms acid with the oil, corroding the windings over time.
Charge according to the nameplate plus the pipe allowance. For a Ø22 liquid line, the allowance is about 0.35 kg/m; 30 m adds 10.5 kg. If the nameplate charge is 45 kg, the actual charge is 55.5 kg. Record ambient temperature, suction and discharge pressures, subcooling and superheat after charging. These become the baseline for a future shortage diagnosis.
Test run: seven measurements
“It cools” is not a test. Measure seven items against the technical agreement.
| Item | Criterion | Common cause of failure |
|---|---|---|
| Chilled-water flow | Rated ±10% | Unflushed pipe, blocked strainer, low pump head |
| Temperature difference | Design ±0.5 K, often 5 K | Excess flow, insufficient load |
| Running current | Not above nameplate rated current | Low voltage, high condensing pressure |
| Measured capacity | Not below the lower tolerance | Condition not corrected |
| COP | Within label tolerance | Fouling, low charge, low water flow |
| Noise | Contract value at 1 m | Failed isolation, uneven base |
| Protection | Flow, high/low pressure, freeze, overload trip | Parameter changed, interlock bypassed |
Site capacity for water is Q = 1.163 × V × ΔT, with V in m³/h and ΔT in K. The example gives 1.163 × 86 × 4.6 = 460 kW. If the condenser water is 32 °C instead of the nominal 30 °C, the manufacturer’s curve may correct the nominal capacity to 500 × (1 − 0.04) = 480 kW. The measured 460 kW is 95.8% of the corrected value, inside the tolerance, so the unit passes. If the contract has no correction method, the 20 kW gap cannot be resolved.
COP is measured capacity divided by measured input power. At 480 kW and 96 kW, COP is 5.0. The controller display is often internally corrected; acceptance should use an external power meter and flow meter.
Part-load check
IPLV is a weighted value at 100%, 75%, 50% and 25% load. A full site test is impractical. Check one or two points, usually 75% and 50%, because those represent the longest operating hours and reveal whether staging control is sound. Reduce load through the terminals or a bypass, stabilize for 30 minutes, then record capacity, input power and chilled-water outlet stability. Two criteria: COP at that point not below the catalog tolerance, and outlet variation within ±0.5 K. The second is often ignored, but a unit swinging 2 K cannot deliver stable comfort or process conditions.
Handover
The value of the handover file is not its thickness but whether someone can understand it three years later. Keep four items as separate volumes: the control parameter table (protection values, alarm values, PID settings), the valve and instrument tag list, the spare-parts list with replacement criteria, and the commissioning baseline data.
The parameter table must record the values at the end of commissioning, not the factory defaults. With it, a later change can be detected. Without it, a mistaken change of the freeze protection to −15 °C can crack a heat exchanger in winter with no traceable cause.
Boundaries
Standard clauses follow the current edition. Tolerances and test pressures should follow the contract and the manufacturer’s file. The numerical examples are typical conditions; a project must use its own measurements and performance curves. A system acceptance also covers pumps, cooling towers, controls and piping, which are outside the chiller capacity test.
Common questions
Can a chiller be accepted without a type test report?
Check the contract, but from a compliance view the file should be completed. A type test proves the model design and performance; a factory test proves only that this unit was assembled correctly. Tenders, efficiency registration and green-factory applications often require the type test.
What pressure drop is acceptable over 24 hours?
Correct for temperature first using p₂ = p₁ × T₂ ÷ T₁ in kelvin. After correction, a drop within 1% is acceptable. Without correction, an 8 K day-night swing changes the reading by about 2.7% and causes false judgments.
Is 8% below nameplate a failure?
Correct the condition first. Condenser water 2 K higher can reduce capacity by 3–4%; chilled-water flow 10% low can inflate the temperature difference and understate capacity. Only after correction, if it is still below the tolerance, is it a failure.
How much difference between label COP and site COP is normal?
Typically 3–8%, from fouling, water-flow deviation and instrument error. Above 10%, check condenser fouling, refrigerant charge and sensor drift before blaming the machine.
What is most often missing at handover?
The control parameter table. Most projects hand over the manual and drawings but not the commissioned protection, alarm and PID values. Without a baseline, a controller replacement or parameter change cannot be verified.
How should a failed acceptance be handled?
Separate installation and commissioning causes from equipment performance. Installation issues are corrected and retested, with retest limited to the failed item and related items. Performance failure is handled under the contract tolerance clause with the original records and calibrated instruments.
What test condition?
The contract condition, with corrected capacity and COP. State the tolerance and the instruments. A site reading without correction is not acceptance.
What documents?
Test report, calibration, drawings, O&M, spare list and warranty. Archive them with the acceptance signature.
Related reading
- Water Chiller Commissioning: Sequence and Measurements
- COP vs IPLV: What the Two Figures Measure
- Water Chiller Installation: mechanical room Layout
- Water Chiller Fault Codes and Troubleshooting
Related reading
COP and IPLV guarantee · Selection margin mistakes · Chiller retrofit payback