COP and IPLV are easy to define and hard to guarantee. COP is the full-load efficiency at a standard rating point. IPLV is a weighted part-load number. A catalog can show a high COP and a high IPLV, but neither tells you what the machine will do at your condenser-water temperature and load profile.
This article is about the guarantee clause: what to measure, at what condition, with what tolerance and what happens if the machine misses.
Two numbers, two jobs
COP is the ratio of cooling capacity to input power at a defined condition, usually 7/12 °C chilled water and 30/35 °C condenser water. IPLV is an integrated part-load value, calculated from efficiencies at 100%, 75%, 50% and 25% load with a weighting. AHRI Standard 550/590 uses a 1/10/30/60 weighting at those four points for water-cooled machines.
A machine with COP 5.5 and IPLV 6.2 is not “better” than one with COP 5.2 and IPLV 6.4 without the load profile. If the plant runs at 80% load for most hours, COP matters more. If it runs at 40–60% for most hours, IPLV and the part-load curve matter more.
| Metric | Condition | Use |
|---|---|---|
| COP | Full load, standard temperatures | Compare machines at design |
| IPLV | Weighted 100/75/50/25% | Compare annual part-load |
| NPLV | Part-load at site temperatures | Guarantee for the project |
| Annual kWh | Hour-by-hour load and ambient | Decide the energy bill |
The guarantee should use NPLV, not catalog IPLV
A standard IPLV assumes 30/35 °C condenser water and a standard fouling factor. A site with a cooling tower may run at 32/37 °C in summer and 24/29 °C in winter. A site with an air-cooled condenser sees a condensing temperature that follows ambient. The catalog IPLV is not the site IPLV.
Ask the supplier to calculate an NPLV at the site’s condenser-water temperatures and the actual load weighting. If the supplier cannot, the guarantee should be based on a measured test at the site, not on a catalog number.
How to write the clause
A workable clause has five parts: the metric, the condition, the measurement method, the tolerance and the remedy. “COP not less than 5.5” is not a clause; it has no condition, no method and no consequence.
| Part | Example wording |
|---|---|
| Metric | Full-load COP and NPLV at site conditions |
| Condition | 7/12 °C chilled water; condenser water per design; fouling factor stated |
| Method | Factory witness test or site test with calibrated instruments; 30 min steady state |
| Tolerance | Not less than 95% of the guaranteed value |
| Remedy | Price adjustment per kWh shortfall, or correction, or replacement |
The tolerance matters. A 5% tolerance on a 5.5 COP is 0.275 COP, which is a real energy difference. Do not accept a 10% tolerance unless the price reflects it.
The money at stake
Take a 500 kW machine running 2,000 full-load-equivalent hours a year. At COP 5.5, input power is 90.9 kW; at COP 5.2, it is 96.2 kW. The difference is 5.3 kW, or 10,600 kWh/year. At CNY 0.70/kWh, that is CNY 7,400/year. Over 10 years, it is CNY 74,000, before the part-load effect.
The part-load effect can be larger. A machine with a flat part-load curve may use 3–5% less energy over a year than a machine with a high full-load COP but a poor 50% point. The guarantee should include both, or the annual energy calculation should be the primary metric.
Test conditions that actually match
The test should use the design chilled-water flow, the design leaving-water temperature, the design condenser-water flow and temperature, and the design fouling factor. A test at a lower condenser temperature flatters the COP. A test at a reduced flow flatters the capacity. A test with a clean condenser flatters the approach.
Record the flow, the temperatures, the power and the barometric pressure. Correct the capacity and power to the guarantee condition using the manufacturer’s correction curves, and attach the raw data. A test report with only a final number is not auditable.
Part-load weighting for the site
Do not use the standard 1/10/30/60 weighting if the site never runs at 25%. A process plant that runs at 70–85% all year has a different weighting. A hotel with a night load of 20% has another. Build the weighting from 12 months of load data, or from a load profile if the plant is new.
The guarantee can use a simpler annual energy cap: “annual energy at the design load profile not more than X kWh”. That is easier to enforce than a single IPLV number, but it needs a agreed load profile and a weather file.
Measurement uncertainty
A single test has uncertainty. A flow meter may be 2% off, a temperature sensor 0.5 K, a power meter 1%. At a 5 K condenser rise, a 0.5 K error is 10% of the rise and therefore a large error in the calculated capacity. The test standard should state the instrument class, the calibration date and the allowed uncertainty.
A guarantee that ignores uncertainty invites a dispute. The clause can define a test with a stated uncertainty and a tolerance that is wider than the uncertainty. If the tolerance is 5% and the uncertainty is 3%, the result is not decisive. Use a factory witness test with calibrated instruments, or a site test with a third-party meter.
Boundary
This clause covers the chiller, not the pumps, the tower fans or the chilled-water distribution. A system guarantee must define the boundary. A chiller that meets COP but runs with a fouled condenser and a throttled pump is not a system that meets the energy target.
Common questions
Which is more important, COP or IPLV?
Neither without the load profile. COP is the design point; IPLV is a weighted part-load value. For a plant that runs at part load, IPLV and the part-load curve matter more.
Can I use the catalog IPLV in a guarantee?
Only if the site condenser-water temperatures match the rating. Otherwise ask for an NPLV at site conditions, or a site test with calibrated instruments.
What tolerance is reasonable?
About 5% on COP. A 10% tolerance gives away a real energy difference. The remedy should be a price adjustment or correction, not a verbal promise.
How is IPLV weighted?
AHRI 550/590 uses 1/10/30/60 at 100/75/50/25% load for water-cooled machines. A site with a different load profile should use its own weighting.
What instruments are needed?
Calibrated flow meter, temperature sensors, power meter and a pressure gauge. Record the raw data, not only the final COP.
Does the guarantee cover the whole system?
No, unless the contract says so. The chiller guarantee covers the chiller. Pumps, tower fans and distribution need a separate boundary and a separate target.
What tolerance is normal?
A guarantee should state the test standard, the site condition and a tolerance, often 3–5%. A number without a tolerance cannot be enforced. The remedy should be repair, re-test or a price adjustment.
How is NPLV tested?
Use the applicable standard, the specified part-load points and the site conditions. The test must be witnessed or certified. A catalog curve is not a test report.
Further technical reading
For the definitions, see COP vs IPLV: What the Two Efficiency Figures Actually Measure. For part-load selection, see Chiller Selection: Full-Load COP Is Not the Decision. For staging, see Why Chillers Are More Efficient at Part Load. For magnetic machines, see Magnetic vs Ordinary Centrifugal: Pay for Hours, Not IPLV.
Related reading
Selection margin mistakes · Chiller plant group control · Chiller retrofit payback