Three quotations for the same 30 kW heat-pump plant can differ by 40%. The difference is rarely profit. One supplier makes the chassis, coils and controls; one buys the parts and assembles; one forward a price from another factory. All three can provide a glossy catalog. Only one can tell you which heat exchanger will be available in year eight.
The audit is not a search for the biggest name. It is a check of identity, model-specific evidence and low-temperature performance. This article sets out the four factory checkpoints, the six documents, the -7 °C witness test and the five contract clauses that make the audit useful.
First, establish the identity
Ask for the production address and a list of in-house processes. A manufacturer with a fin press, tube expander, bending line, sheet-metal paint line and an integrated performance test bench is a real manufacturer. A supplier with only an assembly line is an assembler. A supplier that cannot name a production address is a trader.
None of these identities is automatically bad. A trader is convenient for a small or urgent order. For a project above CNY 1 million, however, the service and spare-parts chain matters more than the first price. The audit should establish which identity you are dealing with before you compare the numbers.
Four factory checkpoints
| Checkpoint | What good looks like | What to question |
|---|---|---|
| Performance test bench | Integrated enthalpy lab; range covers the offered model; each unit logged | Component-only test or third-party report for a different size |
| Coil production | Fin press, tube expander and bender in-house | All coils bought in; only assembly on site |
| Work in progress | The offered model is in production or in stock | Only small units visible; the large unit is “made after order” |
| Loaded run area | At least 4 hours of loaded run before dispatch, with a record | Power-on test only, no water loop or load |
Two hours on the floor are enough to see these four points. Look at the test-bench nameplate range, not the brochure. A bench that stops at 20 kW cannot witness a 40 kW unit at full load. Ask for the last three dispatch records of the same model and see whether the capacity, power and water temperatures were recorded or only a pass/fail box was ticked.
Six documents, checked by number
Business license and scope. Check the national enterprise credit system. A scope that does not include refrigeration equipment manufacturing is a warning, not a disqualification; the supplier may be a trader.
ISO 9001. Check the certificate number and expiry on the certification body’s site. A certificate that covers “sales” but not “design and manufacture” does not cover the factory.
CCC certificate. CCC is issued by model. A certificate for a 5 kW unit does not cover a 40 kW unit. Read the annex and find the offered model in the list.
Efficiency report. The sample model must match the offered model. A report for a smaller unit in the same “series” is not a report for the unit you are buying.
Low-temperature report. Ask for measured data at -7 °C or -12 °C: heating capacity, input power and COP. A nominal-condition report plus a “correction curve” is not a witness test.
Patents and software copyrights. Useful context, not proof of manufacturing capability. An appearance patent does not tell you whether the coil can be replaced.
The -7 °C witness test
The most useful test is not at 7 °C. It is at the ambient condition where the project actually needs heat. For a cold-climate project, witness at -7 °C with the design leaving-water temperature, and again at -12 °C if the site requires it.
| Test point | Record | Why it matters |
|---|---|---|
| Nominal capacity | Heating capacity, input power, COP at 7 °C | Baseline for the catalog number |
| Low-temperature capacity | Capacity, power, COP at -7 °C | Defines the real unit size for winter |
| Defrost | Defrost interval, duration, capacity recovery | A unit can pass a steady test and fail in cycling |
| Leaving water | Temperature at design flow, not reduced flow | Low flow flatters the capacity number |
A 30 kW nominal unit may deliver 18–22 kW at -7 °C, depending on compressor, coil and refrigerant. That is not a fault; it is physics. The fault is buying 30 kW of nominal capacity for a 25 kW winter load without asking for the -7 °C number. The witness test should use the design water flow and the design leaving-water temperature, and it should record the defrost cycle.
References: three items, not a photo album
Ask for the contract page showing the equipment list and model, the acceptance certificate with the owner’s stamp, and a site photo showing the unit nameplate. A photo of a hotel mechanical room proves a hotel exists; it does not prove the same model was installed or that it works.
Ask three questions: Is the project still running? What outlet temperature does it actually hold? What failed in the first two years? A supplier who answers with numbers and names has a service record. A supplier who answers “all projects are good” has a brochure. Reference similarity matters more than reference count: a school dormitory and a five-star hotel have different draw profiles, installation constraints and maintenance teams.
Five contract clauses
| Clause | Write this | If omitted |
|---|---|---|
| Model and configuration | Compressor brand and model; coil type, rows and fin spacing; controller model | A substitute arrives and the price looks the same |
| Performance | Test condition, flow, leaving-water temperature and remedy if short | Dispute after the first winter |
| Spare parts | At least 8 years after discontinuation; key parts price list | No coil available in year six |
| Service | On-site response time, nearest parts stock, named technician | “24-hour response” becomes a call center |
| Warranty | Unit, compressor and controller listed separately; water quality exclusions stated | Year-two failure blamed on the site |
The first clause is the one most often softened. “High-efficiency scroll compressor” is not a specification. Write the brand and model. “High-efficiency heat exchanger” is not a specification. Write brazed plate or shell-and-tube, the surface area and the material. These details do not change the sales story; they decide the repair bill in year five.
Boundary
A full factory audit is not economic for a CNY 300,000 project. For small projects, concentrate on the model-specific reports and the contract clauses. For imported brands assembled locally, check the local spare-parts stock rather than the foreign factory. When buying through an engineering contractor, audit the manufacturer but sign with the contractor; confirm the manufacturer’s authorization and the responsibility boundary in writing.
Common questions
How do I tell a manufacturer from an assembler?
Ask for the production address and the in-house process list. Fin press, tube expander, bender, paint line and an integrated test bench indicate manufacturing. An assembly line only indicates assembly; no address indicates a trader.
Which certificate is most often mismatched?
CCC and the efficiency report. CCC is issued by model, so a certificate for a small unit does not cover a large one. The efficiency report sample model must match the offered model, including any suffix.
Why test at -7 °C?
Because the nominal 7 °C capacity is not the winter capacity. A 30 kW nominal unit may deliver 18–22 kW at -7 °C. The project should be sized on the low-temperature number, not the catalog headline.
Is a 4-hour loaded run enough?
It is a practical minimum for a factory witness test. It catches assembly faults, control faults and refrigerant leaks. It does not prove five-year reliability; that requires reference sites and spare-parts evidence.
What should the contract say about spare parts?
At least eight years of supply after discontinuation, with a price list for the coil, compressor, controller and fans. Without it, a working unit can become scrap for want of one part.
Does a lower price always mean a trader?
No. It can mean a smaller coil, a different compressor, a thinner frame or a shorter warranty. Compare the configuration line by line before comparing the total.
What test point?
-7 °C ambient with the design water flow. A paper curve at 7 °C does not prove cold-climate capacity.
What data?
Capacity, power, defrost time and supply temperature. Record them at the same time. A single number is not a test.
Further technical reading
For the heat-pump audit, see Heat-Pump Factory Audit. For payback, see Heat Pump Payback. For installation, see Heat-Pump Installation: Six Stages, Three Inspections. For service, see Air-Source Heat Pump Maintenance.
Related reading
Heat Pump Factory Audit Low Temperature · Heat Pump COP Rating · Heat Pump Tender Technical Response